WO2012096175A1 - 行動パタン解析装置、行動パタン解析方法および行動パタン解析プログラム - Google Patents
行動パタン解析装置、行動パタン解析方法および行動パタン解析プログラム Download PDFInfo
- Publication number
- WO2012096175A1 WO2012096175A1 PCT/JP2012/000140 JP2012000140W WO2012096175A1 WO 2012096175 A1 WO2012096175 A1 WO 2012096175A1 JP 2012000140 W JP2012000140 W JP 2012000140W WO 2012096175 A1 WO2012096175 A1 WO 2012096175A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- position information
- time
- positioning
- information log
- log
- Prior art date
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/284—Relational databases
- G06F16/285—Clustering or classification
- G06F16/287—Visualization; Browsing
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S19/00—Satellite radio beacon positioning systems; Determining position, velocity or attitude using signals transmitted by such systems
- G01S19/01—Satellite radio beacon positioning systems transmitting time-stamped messages, e.g. GPS [Global Positioning System], GLONASS [Global Orbiting Navigation Satellite System] or GALILEO
- G01S19/13—Receivers
- G01S19/24—Acquisition or tracking or demodulation of signals transmitted by the system
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S19/00—Satellite radio beacon positioning systems; Determining position, velocity or attitude using signals transmitted by such systems
- G01S19/01—Satellite radio beacon positioning systems transmitting time-stamped messages, e.g. GPS [Global Positioning System], GLONASS [Global Orbiting Navigation Satellite System] or GALILEO
- G01S19/13—Receivers
- G01S19/35—Constructional details or hardware or software details of the signal processing chain
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S19/00—Satellite radio beacon positioning systems; Determining position, velocity or attitude using signals transmitted by such systems
- G01S19/01—Satellite radio beacon positioning systems transmitting time-stamped messages, e.g. GPS [Global Positioning System], GLONASS [Global Orbiting Navigation Satellite System] or GALILEO
- G01S19/13—Receivers
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S19/00—Satellite radio beacon positioning systems; Determining position, velocity or attitude using signals transmitted by such systems
- G01S19/38—Determining a navigation solution using signals transmitted by a satellite radio beacon positioning system
- G01S19/39—Determining a navigation solution using signals transmitted by a satellite radio beacon positioning system the satellite radio beacon positioning system transmitting time-stamped messages, e.g. GPS [Global Positioning System], GLONASS [Global Orbiting Navigation Satellite System] or GALILEO
- G01S19/42—Determining position
Definitions
- the present invention relates to an action pattern analysis device, an action pattern analysis method and an action pattern analysis program for analyzing an action pattern from position information measured irregularly.
- LBS Location Based Service
- an application that uses the GPS (Global Positioning System) function installed in mobile phones etc. as LBS and an application that adds location information logs to photos taken with a mobile terminal with a camera function, and "Tweet" with location information logs Twitter (registered trademark) is widely used.
- LBS there is a recommendation service that performs content distribution in accordance with a user's action pattern that can be obtained from personal position information.
- Patent Document 1 describes an action management device capable of managing an action of an individual from position information.
- the behavior management device described in Patent Document 1 searches for information for performing behavior management and estimates behavior from position information measured at predetermined time intervals using a GPS function.
- the positioning interval of the position information using the terminal with the position information acquisition function may be long, or the positioning of the position information may be irregular.
- the behavior management device described in Patent Document 1 has a problem that the behavior management can not be performed with high accuracy.
- the positioning interval of the position information is long and the positioning of the position information is irregular, the location where the user often stays and the stay time zone, the moving route and the moving time where the user frequently moves It is desirable that the user's daily behavior patterns, such as bands, be accurately obtained.
- the present invention provides an action pattern analysis device, an action pattern analysis method, and an action that can analyze the action pattern of the user with high accuracy even when the positioning interval of the position information is long and positioning of the position information is irregular.
- the purpose is to provide a pattern analysis program.
- a behavioral pattern analysis device plots a positional information log, which is information including a user's positioning position and positioning date and time, in a multidimensional space defined by numerical information representing the positioning position and time.
- the position information log is weighted so that it is easily determined that the Euclidean distance in the time direction with respect to the position information space, which is a space defined by the numerical value information representing the positioning position in the multidimensional space
- a retention point cluster extraction unit that extracts a retention point that is a position at which a user frequently stays by clustering information logs, and a position information log extracted as a retention point from the position information log plotted by the position information plotting unit
- Non-dwelling point positional information log extracting means for extracting a set of positional information logs excluding a point as a non-dwelling point
- the moving route of the user is extracted by performing weighting on the position information log of the non-staying point, which makes it easy to determine that the Euclidean distance in the time direction to the position information space is
- Another behavior pattern analysis device is a difference value between two adjacent position information logs sorted in the order of positioning date and time included in the position information log which is information including the user's positioning position and positioning date and time.
- the attribute of the position information log is based on the time difference between the positioning dates and times included in the position information log, the movement vector function value calculating means for calculating the distance between the positioning positions, and the time difference between the positioning dates and times and the distance between the positioning positions
- Attribute determination means for determining whether a retention attribute indicates a retention point at which the user frequently stagnates or a non-retention attribute indicates a non-retention point at a position on the user's travel route;
- a position information plotting means for plotting on a multidimensional space defined by numerical information representing the positioning position and time, and an sky defined by the numerical information representing the positioning position in the multidimensional space
- the position information log is weighted so that it is easily determined that the Euclidean distance in the time direction to the position information space is short, and the position information log
- a behavioral pattern analysis method plots a position information log, which is information including a user's positioning position and positioning date and time, in a multidimensional space defined by numerical information representing the positioning position and time, and multidimensional A weighting is applied to the position information log to easily determine that the Euclidean distance in the time direction is close to the position information space which is a space defined by the numerical information representing the positioning position in space, and the weighted position information log is clustered.
- the retention point which is the position where the user frequently dwells
- a set of position information logs excluding the position information log extracted as the retention point from the position information log plotted on the multidimensional space is not It is extracted as a retention point, and it is not possible to determine the weighting that makes it easy to determine that the Euclidean distance in the time direction to the position information space is far Performed on the position information log boiling point, by clustering the weighted location information log, and extracts the movement path of the user.
- Another action pattern analysis method is a difference value between two adjacent position information logs sorted in the order of positioning date and time included in the position information log which is information including the user's positioning position and positioning date and time.
- the time difference between positioning dates and times included in the position information log and the distance between positioning positions are calculated, and the attribute of the position information log is frequently stagnated by the user based on the time difference between positioning dates and times and the distance between positioning positions It is determined whether it is a staying attribute that indicates the staying point that is the position to be located or a non-dwelling attribute that indicates the non-dwelling point that is the position on the user's moving route, and the position information log is defined by numerical information And the Euclidean distance in the time direction to the position information space, which is a space defined by numerical information representing the positioning position in the multidimensional space, is determined to be close.
- Weighting is performed on the position information log, and by clustering the position information logs determined to have the holding attribute among the weighted position information logs, the holding point is extracted, and the time direction relative to the position information space is extracted.
- the moving route of the user is performed by performing weighting on the position information log that makes it easy to determine that the Euclidean distance is far, and clustering the position information logs determined to have the non-staying attribute among the weighted position information logs. To extract.
- a behavioral pattern analysis program plots on a computer a position information log, which is information including a user's positioning position and positioning date and time, in a multidimensional space defined by numerical information representing the positioning position and time.
- Position information plotting processing, weighting is applied to the position information log to make it easy to determine that the Euclidean distance in the time direction is close to the position information space which is a space defined by numerical information representing a positioning position in a multidimensional space.
- a retention point cluster extraction process for extracting retention points, which are locations where users frequently stay, by clustering the position information logs, and a position extracted as a retention point from the position information log plotted in the position information plot process A non-dwelling point for extracting a set of position information logs excluding the information log as a non-dwelling point
- Another behavioral pattern analysis program is a computer configured to compare the difference between two adjacent position information logs sorted in the order of positioning date and time included in the position information log which is information including the user's positioning position and positioning date and time. Based on the time difference between positioning dates and times included in the position information log as a value, movement vector function value calculation processing for calculating the distance between positioning positions, the time difference between positioning dates and times, and the distance between positioning positions, Attribute determination processing to determine whether the attribute is a residence attribute indicating a residence point where the user frequently resides, or a non-dwelling attribute indicating a non-dwelling point that is a position on the user's travel route, position information log Position information plotting processing for plotting on a multi-dimensional space defined by numerical information representing the positioning position and time, to numerical information representing the positioning position in the multi-dimensional space A position where the Euclidean distance in the time direction with respect to the position information space, which is a space defined as described above, is determined to be close to the position information log,
- the user's action pattern can be analyzed with high accuracy.
- the life log service is such that location information and acquisition time of location information are recorded using sensor information based on GPS or the like mounted on a portable terminal.
- FIG. 1 is a block diagram showing a configuration example of a first embodiment of the behavior pattern analysis device according to the present invention.
- the behavior pattern analysis device in the present embodiment includes a position information log input unit 10, a behavior pattern analysis reference input unit 11, a behavior pattern analysis unit 20, and a behavior pattern storage unit 30.
- the position information log is information including numerical value information indicating the position of a terminal that has acquired a position such as latitude, longitude, and altitude, and date and time information indicating the date and time when the numerical value information (position information) is acquired.
- the position information log is information including the positioning position of the terminal user and the positioning date and time.
- the position information log may include an identifier of a terminal that has acquired the position information, and an identifier of the terminal user (hereinafter referred to as a user identifier).
- FIG. 2 is an explanatory view showing an example of the position information log. The example illustrated in FIG. 2 indicates that the location information log includes the user identifier, the date and time, the latitude and the longitude.
- the action pattern analysis reference input unit 11 performs action information (hereinafter, also simply referred to as a reference) as a reference when the action pattern analysis unit 20 described later analyzes the action pattern based on the position information log. Input to the pattern analysis unit 20.
- the behavior pattern analysis standard input unit 11 is a standard of the day to be analyzed (such as "Thursday", "all days") (hereinafter referred to as a day standard) or "3 am"
- the reference of the time to be an analysis target such as (hereinafter referred to as a time reference) is input to the behavior pattern analysis unit 20. That is, the reference here is information specifying the date and time information of the position information log to be analyzed.
- the action pattern analysis unit 20 is a position information where a terminal user frequently stays (hereinafter referred to as a staying point) and a route where the terminal user frequently travels (hereinafter referred to as a movement route) are position information. Extract by analyzing the log. Hereinafter, these extracted information may be collectively referred to as action pattern data.
- the action pattern analysis unit 20 includes a clustering space plotting unit 21, a staying point cluster extracting unit 22, a staying point position information log removing unit 23, and a movement path cluster extracting unit 24.
- the clustering space plotting unit 21 plots location information logs to be analyzed in a clustering space.
- the clustering space means a super space in which a time axis representing time is added to a numerical value representing a position such as latitude, longitude, or altitude.
- the clustering space is a multidimensional space defined by position information and date and time information, and is a three-dimensional space defined by at least latitude, longitude, and time.
- plotting the position information log in the clustering space means indicating the position information log in the coordinates of the clustering space.
- the positional information log plotted in the clustering space may be referred to as data plotted in the clustering space or clustering spatial data.
- FIG. 3 is an explanatory view showing an example of a clustering space.
- the clustering space illustrated in FIG. 3 is a three-dimensional space defined by the latitude axis, the longitude axis, and the time axis, and corresponds to the x axis, the y axis, and the z axis, respectively. The details of FIG. 3 will be described later.
- the clustering space plotting unit 21 holds the day of the week reference and the time reference value input from the action pattern analysis reference input unit 11. The clustering space plotting unit 21 then performs position information logging in the clustering space based on the held day of the week reference and time reference values, and the position information log of the user input from the position information log input unit 10. Plot The clustering space plotting unit 21 sets, for example, the position information log acquired within a predetermined period from the reference when analyzing the action pattern as an analysis target. Thereafter, the clustering space plotting unit 21 inputs the data plotted in the clustering space to the staying point cluster extracting unit 22.
- FIG. 4 is an explanatory view showing an example of the position information log plotted in the clustering space.
- FIG. 4 shows an example where position information logs are plotted in a clustering space in which a latitude-longitude plane is divided for each date, and the position information logs are integrated.
- the clustering space plotting unit 21 is a position information log from Thursday, September 2 from 3 am to September 3, 3 am, and September 9, Thursday from 3 am to September 10, 3 am Extract location information logs up to Then, the clustering space plotting unit 21 removes the date of the extracted position information log, and adds the time axis aligned on the time series from 3 am to 3 am the next day to the clustering space where the time axis is added to the latitude-longitude plane The position information log is plotted (see FIGS. 4 (a) and (b)). Then, the clustering space plotting unit 21 generates a clustering space in which the position information logs for these two days are integrated (see FIG. 4C).
- the clustering space plotting unit 21 plots position information logs for each date on a two-dimensional plane defined by latitude and longitude, and then integrates the position information logs.
- the clustering space plotting unit 21 may plot the position information log directly in a three-dimensional clustering space defined by latitude, longitude and time.
- the clustering space plotting unit 21 plots the position information log in a four-dimensional clustering space defined by the latitude, longitude, altitude and time. You may
- the staying point cluster extraction unit 22 performs clustering of clustering spatial data, and extracts a position where the terminal user is staying frequently (that is, a staying point).
- the staying point cluster extraction unit 22 is a general-purpose generalization such as EM (Expectation Maximization) algorithm or hierarchical clustering, which is a set of position information logs having close Euclidean distance in space consisting of three variables of latitude, longitude and time.
- EM Extractation Maximization
- hierarchical clustering which is a set of position information logs having close Euclidean distance in space consisting of three variables of latitude, longitude and time.
- the dwell point is extracted by using the clustering method of
- the extracted retention points may be referred to as retention point pattern data.
- the space in which the staying point cluster extraction unit 22 performs clustering is not limited to a space defined by three variables of latitude, longitude, and time.
- the space for clustering may be a space defined by four or more variables.
- the staying point cluster extraction unit 22 performs clustering on the location information log in the space defined by the four variables of latitude, longitude, altitude, and time. It is also good.
- the position information log is plotted on the space defined by the three variables of latitude, longitude and time will be specifically described.
- the position information log the latitude at the coordinates indicated by X i, longitude Y i, if the time and Z i, 2 squared value of the Euclidean distance between certain two points, (X 1 -X 2) 2 + ( It is defined by Y 1 -Y 2 ) 2 + (Z 1 -Z 2 ) 2 .
- time Z is different in unit system. Therefore, it is necessary to define a distance function that is strongly related to the location (latitude-longitude) with respect to the time in space. Therefore, a distance function is defined in which the value of (Z 1 -Z 2 ) 2 is multiplied by a predetermined weight value k 1 . That is, this function is defined as (X 1 -X 2 ) 2 + (Y 1 -Y 2 ) 2 + k 1 (Z 1 -Z 2 ) 2 .
- the weight value k 1 is the difference between the distance between the points on the latitude-longitude plane and the distance between the points on the latitude-longitude plane and the points along the time axis. It can be said that it is a weighting value to be attached.
- the value of k 1 is decreased, when the position information logs are compared, the difference in time becomes smaller than the difference in distance. Therefore, it is easy to determine that the Euclidean distance of the time axis with respect to the latitude-longitude plane is close by reducing the value of k 1 , so in the position information log, a log close to a place (that is, a position determined from latitude and longitude) It is possible to generate clusters that are likely to be in the same set.
- This weight value k 1 the result that the user has been previously tuned so that the size of the intended retention point of service is specified. For example, when time Z is an hour unit, when it is desired to extract a staying point having a radius of about 1 km in Japan, it is desirable to set the weight value k 1 to about 1/1200.
- the weight values set in advance are compared with the latitude and longitude in the data plotted in the clustering space. And multiply each.
- the weight value may be any value as long as it is determined that the Euclidean distance on the time axis is close to the latitude-longitude plane.
- the retention point cluster extraction unit 22 performs a clustering operation on the clustering spatial data in order to extract the retention points using the clustering spatial data multiplied by the weight value.
- the operation of clustering may be performed using any generally known clustering method.
- the case where the clustering operation is performed using the centroid method will be described as an example.
- the retention point cluster extraction unit 22 sets all data plotted in the clustering space as initial clusters, and integrates clusters having the shortest Euclidean distance between two points selected from all clusters into one cluster. At this time, the staying point cluster extraction unit 22 obtains the value of the center of gravity from the value included in the original cluster, and sets the obtained value as a new data value of the integrated cluster. The retention point cluster extraction unit 22 also holds the number of clustering spatial data forming the integrated cluster.
- the staying point cluster extraction unit 22 repeatedly performs the above calculation within a range in which the Euclidean distance does not exceed a preset threshold (hereinafter, referred to as a first Euclidean distance threshold).
- a preset threshold hereinafter, referred to as a first Euclidean distance threshold.
- the staying point cluster extraction unit 22 determines a cluster whose number of clustering spatial data obtained by the operation up to this point exceeds a preset threshold (hereinafter referred to as a first cluster number threshold) as the staying point.
- a preset threshold hereinafter referred to as a first cluster number threshold
- An appropriate value is set to this first cluster number threshold value by a test or the like.
- the position at which the position information of the number exceeding the first cluster number threshold value can be said to be the position where the user frequently stays.
- the retention point cluster extraction unit 22 determines that the clusters 301a and 301b are retention points. Then, the staying point cluster extraction unit 22 identifies the staying point of the earliest time and the latest time of the data in the cluster, and the value of the center of gravity of the position information (latitude and longitude) in the data in the cluster.
- the identifier is stored in the retention point pattern storage unit 31 together with an identifier (hereinafter, referred to as a retention point identifier).
- the staying point position information log removing unit 23 extracts, from the data passed from the clustering space plotting unit 21, a position information log that the staying point cluster extracting unit 22 has not determined as a staying point.
- the set of position information logs is referred to as a non-staying point.
- the movement path cluster extraction unit 24 performs clustering of clustering spatial data, and extracts a time zone and movement path in which the terminal user is moving frequently.
- the extracted travel route may be referred to as travel route pattern data.
- the movement route cluster extraction unit 24 generates movement by using a general-purpose clustering method to generate a set of position information logs having close Euclidean distance in space consisting of three variables of latitude, longitude, and time. Extract the route.
- the space where the movement path cluster extraction unit 24 performs clustering is not limited to the space defined by the three variables of latitude, longitude, and time, but is defined by four or more variables, as described in the retention point cluster extraction unit 22. It may be a space to be
- the distance function for the three variables of latitude, longitude and time is (X 1 -X 2 ) 2 + (Y 1 -Y 2 ) 2 + k 2 (Z 1 -Z) 2) 2 and is defined, latitude, longitude, distance function for 4 variable altitude and time, (X 1 -X 2) 2 + (Y 1 -Y 2) 2 + (W 1 -W 2) 2 + k 2 It is defined as (Z 1 -Z 2 ) 2 .
- Weight value k 2 Here also, as a result of the user has been pre-tuned so that the size of the intended retention point of service is specified.
- the relationship of k 1 ⁇ k 2 is established between the weight value k 1 when clustering the staying points and the weight value k 2 when calculating the movement path.
- the movement path cluster extraction unit 24 first receives clustering spatial data that has not been determined as a retention point from among data plotted in the clustering space from the retention point position information log removal unit 23. Then, the movement path cluster extraction unit 24 multiplies the latitude and the longitude of the data plotted in the clustering space with the weight value set in advance.
- the weight may be any value as long as it is determined that the Euclidean distance on the time axis is far from the latitude-longitude plane.
- the movement path cluster extraction unit 24 performs a clustering operation on the clustering space data in order to extract a place indicating the movement path using the clustering space data multiplied by the weight value.
- the operation of clustering may be performed using any generally known clustering method.
- the case where the clustering operation is performed using the centroid method will be described as an example.
- the movement path cluster extraction unit 24 sets clustering space data received from the retention point position information log removal unit 23 as initial clusters, and selects one cluster among the clusters having the shortest Euclidean distance between two points selected from all clusters. Consolidate into two clusters. At this time, the movement path cluster extraction unit 24 obtains the value of the center of gravity from the value included in the original cluster, and sets the obtained value as a new data value of the integrated cluster. The movement route cluster extraction unit 24 also holds the number of clustering spatial data forming the integrated cluster.
- the movement path cluster extraction unit 24 repeatedly performs the above-described calculation in a range in which the Euclidean distance does not exceed a preset threshold (hereinafter referred to as a second Euclidean distance threshold). In addition, what is necessary is just to set a value comparable to a weight value to a 2nd Euclidean distance threshold value.
- the movement path cluster extraction unit 24 determines a cluster whose number of clustering spatial data obtained by the operation up to this point exceeds a preset threshold (hereinafter referred to as a second cluster number threshold) as a movement path.
- a preset threshold hereinafter referred to as a second cluster number threshold
- An appropriate value is set to this second cluster number threshold value by a test or the like. For example, in the case of clustering space data illustrated in FIG. 4C, the movement route cluster extraction unit 24 determines that the cluster 302 is a movement route.
- the movement path cluster extraction unit 24 extracts the earliest time and the latest time of the data in the cluster. Furthermore, the movement path cluster extraction unit 24 determines the clustering spatial data (that is, the value of the center of gravity of the cluster) included in the cluster and the position information of all the retention points stored by the retention point pattern storage unit 31 (that is, A dwell point with the shortest Euclidean distance between two points with and a center of gravity value of longitude and a dwell point second shortest are determined as the start and end points of the movement path. Then, the movement route cluster extraction unit 24 stores the extracted time and start and end points and the determined retention point in the movement route pattern storage unit 32 together with an identifier for identifying the movement route (hereinafter referred to as a movement route identifier). Let Hereinafter, information for identifying the staying point with the shortest Euclidean distance and the staying point with the second shortest distance will be referred to as a first adjacent staying point identifier and a second adjacent staying point identifier.
- the clusters 101a, 101b, and 101c indicate the clusters determined to be retention points, and the clusters 102a and 102b indicate the clusters determined to be movement paths.
- the action pattern analysis unit 20 (more specifically, the clustering space plot unit 21, the retention point cluster extraction unit 22, the retention point position information log removal unit 23, and the movement path cluster extraction unit 24) It is realized by the CPU of the computer operating according to the analysis program).
- the program is stored in a storage unit (not shown) of the behavior pattern analysis device, and the CPU reads the program, and according to the program, the behavior pattern analysis unit 20 (more specifically, the clustering space plotting unit 21, A retention point cluster extraction unit 22, a retention point position information log removal unit 23, and a movement route cluster extraction unit 24 may be operated.
- the clustering space plotting unit 21, the staying point cluster extracting unit 22, the staying point position information log removing unit 23, and the movement path cluster extracting unit 24 may be respectively realized by dedicated hardware.
- the action pattern storage unit 30 stores the action pattern data analyzed by the action pattern analysis unit 20 as a database.
- the action pattern storage unit 30 includes a retention point pattern storage unit 31 and a movement route pattern storage unit 32.
- the retention point pattern storage unit 31 stores the retention point pattern data extracted by the retention point cluster extraction unit 22 as a database.
- FIG. 5 is an explanatory view showing an example of staying point pattern data.
- the retention point pattern storage unit 31 stores the user identifier, the retention point identifier, the retention start time, the retention end time, the latitude and the longitude as retention point pattern data.
- the position information log includes information indicating the altitude
- the retention point pattern storage unit 31 stores retention point pattern data to which the altitude is added.
- the movement route pattern storage unit 32 stores the movement route pattern data extracted by the movement route cluster extraction unit 24 as a database.
- FIG. 6 is an explanatory view showing an example of movement route pattern data.
- the moving route pattern storage unit 32 includes a user identifier, a moving route identifier, a moving start time, a moving end time, a first proximity staying point identifier, and a second proximity staying point identifier as moving route pattern data.
- the action pattern storage unit 30 (more specifically, the staying point pattern storage unit 31 and the movement path pattern storage unit 32) is realized by, for example, a magnetic disk or the like.
- FIG. 7 and 8 are flowcharts showing an operation example of the behavior pattern analysis device.
- FIG. 9 is a flowchart showing an example of an operation of setting a reference day and a reference time.
- the position information log input unit 10 inputs the position information log to the clustering space plotting unit 21 (step A1 in FIG. 7).
- the clustering space plotting unit 21 plots the position information log in the clustering space based on the reference day of the week and the reference time already set, and retains the data plotted in the clustering space Input to the point cluster extraction unit 22 (step A2). The method of setting the reference day and the reference time will be described later.
- the retention point cluster extraction unit 22 receives the data plotted in the clustering space from the clustering space plotting unit 21, the weight values set in advance are compared with the latitude and longitude of the data plotted in the clustering space. And multiply each (step A3). Note that the staying point cluster extraction unit 22 may multiply the time by a preset weight value.
- the staying point cluster extraction unit 22 plots all the data plotted in the clustering space and multiplied by the weight value as an initial cluster (step A4). Then, the staying point cluster extraction unit 22 determines whether the Euclidean distance between two points selected out of all the clusters exceeds the set first Euclidean distance threshold (step A5). If the Euclidean distance between the two points does not exceed the first Euclidean distance threshold (No in step A5), the process proceeds to step A6. If it exceeds (Yes in step A5), the process proceeds to step A7.
- the staying point cluster extraction unit 22 integrates clusters having the shortest Euclidean distance between two points selected from among all clusters into one cluster. Do. At this time, the retention point cluster extraction unit 22 integrates the clusters so as to include the time, latitude, and longitude of each position information log that configures the clusters. Furthermore, the staying point cluster extraction unit 22 calculates the value of the gravity center of latitude and longitude from the position information log included in the cluster separately from each position information log, and uses the value as a new cluster data value. . The retention point cluster extraction unit 22 also holds the number of pieces of clustering spatial data included in the integrated cluster (step A6). After the process of step A6, the processes of step A5 and subsequent steps are performed.
- the retention point cluster extraction unit 22 stores the clusters including the clustering spatial data whose number exceeds the set first cluster number threshold. I will judge. Next, the retention point cluster extraction unit 22 sets the earliest time of the data in the cluster as the retention start time, and the latest time as the retention end time. Then, the retention point cluster extraction unit 22 stores the retention start time, the retention end time, and the value of the center of gravity of the position information (latitude and longitude) in the data in the cluster in the retention point pattern storage unit 31 (step A7).
- the retention point cluster extraction unit 22 inputs the data plotted in the clustering space received in step A2 and the retention point pattern data determined in step A7 into the retention point position information log removal unit 23 (step A8).
- the retention point position information log removal unit 23 receives the data plotted in the clustering space and the retention point pattern data from the retention point cluster extraction unit 22, the retention point pattern data is included in the retention point pattern data from the data plotted in the clustering space Subtract the data determined to be the retention point. Then, the staying point position information log removing unit 23 inputs the remaining data subtracted from the data plotted in the clustering space to the movement route cluster extracting unit 24 (step A9 in FIG. 8).
- the movement route cluster extraction unit 24 receives data which is data plotted in the clustering space and from which the position information log determined to be the retention point has been removed from the retention point position information log removal unit 23,
- the set weight value is respectively multiplied to the latitude and longitude of the received data (step A10).
- the movement route cluster extraction unit 24 may multiply the time by a preset weight value.
- the movement path cluster extraction unit 24 plots all the data plotted in the clustering space and multiplied by the weight value in step A10 as an initial cluster (step A11). Then, the movement route cluster extraction unit 24 determines whether or not the Euclidean distance between two points selected out of all the clusters exceeds the set second Euclidean distance threshold (step A12). If the Euclidean distance between the two points does not exceed the second Euclidean distance threshold (No in step A12), the process proceeds to step A13, and if it exceeds (Yes in step A12), the process proceeds to step A14.
- the movement path cluster extraction unit 24 combines clusters having the shortest Euclidean distance between two points selected from among all clusters into one cluster. Do. At this time, the movement path cluster extraction unit 24 integrates the clusters so as to include the time, latitude, and longitude of each position information log that configures the clusters. Further, the movement route cluster extraction unit 24 calculates the value of the gravity center of latitude and longitude from the position information log included in the cluster separately from each position information log, and uses the value as a new cluster data value. . In addition, the movement route cluster extraction unit 24 also holds the number of clustering spatial data included in the integrated cluster (step A13). After the process of step A13, the processes of step A12 and subsequent steps are performed.
- the movement path cluster extraction unit 24 moves the clusters including the clustering spatial data of the number exceeding the set second cluster number threshold. I will judge. Next, the movement route cluster extraction unit 24 sets the earliest time of the data in the cluster as the movement start time, and the latest time as the movement end time. Furthermore, the movement route cluster extraction unit 24 compares the position information (latitude and longitude) of the cluster with the position information (latitude and longitude) of the staying point included in the staying point pattern data received in step A9.
- the movement path cluster extraction unit 24 determines, as the first adjacent staying point identifier and the second adjacent staying point identifier, the staying point having the first Euclidean distance with the position of the cluster and the second shortest staying point, respectively. . Then, the movement route cluster extraction unit 24 causes the movement route pattern storage unit 32 to store the movement start time, the movement end time, the first proximity dwell point identifier, and the second proximity dwell point identifier (step A14).
- Step B2 when the user inputs the values of the reference day and the reference time, the behavior pattern analysis reference input unit 11 inputs the values of the input reference day and the reference time into the clustering space plotting unit 21 (see FIG. Step B1 in 9).
- the clustering space data plotting unit 21 holds the values of the reference day and the reference time in the clustering space data plotting unit 21.
- the clustering space plotting unit 21 plots the position information log on the three-dimensional space defined by at least the latitude, the longitude, and the time. Then, the retention point cluster extraction unit 22 performs weighting on the position information log so that it is easily determined that the Euclidean distance in the time direction with respect to the latitude-longitude plane is close, and the retention point is extracted by clustering the position information log. Do. Further, the staying point position information log removing unit 23 extracts the non-staying point excluding the position information log extracted as the staying point from the position information log plotted in the space.
- the movement route cluster extraction unit 24 performs weighting on the position information log of the non-staying point by making it easy to determine that the Euclidean distance in the time direction with respect to the latitude-longitude plane is far, and clustering the position information log. , Extract the user's travel route. With such a configuration, even when the positioning interval of the position information is long and the positioning of the position information is irregular, the user's action pattern can be analyzed with high accuracy.
- FIG. 10 is a block diagram showing a configuration example of a second embodiment of the behavior pattern analysis device according to the present invention.
- the behavior pattern analysis apparatus in the present embodiment includes a position information log input unit 10, a behavior pattern analysis reference input unit 11, a behavior pattern analysis unit 50, a behavior pattern storage unit 30, and a movement vector calculation analysis unit 40. ing.
- the contents of the action pattern analysis reference input unit 11 and the action pattern storage unit 30 are the same as in the first embodiment, and thus the description thereof is omitted.
- the positional information log input unit 10 When the positional information log input unit 10 receives a positional information log having [latitude, longitude, time] as one record, the positional information log input unit 10 inputs the positional information log to the movement vector function value calculating unit 41.
- the time can be referred to as a positioning date and time, and the latitude and longitude can be referred to as a positioning location.
- the movement vector operation analysis unit 40 includes a movement vector function value operation unit 41 and a staying / movement route attribute determination unit 42.
- the movement vector function value operation unit 41 sorts the records of the input position information log in chronological order of time included in the position information log. Then, the movement vector function value calculation unit 41 calculates difference values before and after the records sorted in time series order as a vector function value.
- the vector function value may be referred to as a movement vector function value.
- before and after the record means two adjacent records when the records of the position information log are sorted.
- the difference value includes the time difference between the positioning dates and times included in the position information log, and the distance between the positioning locations.
- the difference value may include a traveling direction angle representing a traveling direction between positioning locations.
- the vector function value can be defined as (t 1 , t 2 , r, ⁇ ).
- t 1 indicates the time when the record was recorded.
- t 2 represents a difference (time difference positioning time) from the time when the record was recorded to the time that the next record is recorded.
- r indicates the distance from the position where the record is recorded to the position where the next record is recorded (the distance between positioning locations).
- ⁇ indicates a traveling direction angle representing a traveling direction from the position where the record is recorded to the position where the next record is recorded.
- FIG. 11 is an explanatory view showing an example of movement vector function values.
- the example shown in FIG. 11 indicates that there are positional information logs recorded at 8:30, 9:00, 9:15, 9:30, 10:30 and 11:30 in chronological order.
- the movement vector function value calculation unit 41 calculates the vector function value of the position information log recorded at 11:30 as (9, 0.25, 10, 315 °).
- the unit of time and the difference of time is "hour”.
- the movement vector function value calculation unit 41 may calculate the values of r and ⁇ based on the above-described equation. In the position information log recorded at 11:30, the vector function value for the position information log recorded at 11:30 is not calculated because there is no record recorded next.
- each record is a record representing a residence, a record representing a movement route, or a residence and movement It is determined which of the routes represents a record that can not be determined.
- the record representing the retention means the record included in the retention point
- the record representing the movement path means the record included in the non-retention point.
- an attribute of a record representing retention is referred to as a retention attribute.
- an attribute of a record representing a moving route is indicated as a moving attribute.
- the staying / moving route attribute determining unit 42 determines the attribute of each record indicating the vector function value received from the moving vector function value calculating unit 41, the staying point allowable distance R set in advance, and the effective record threshold time Determine using T.
- the staying / moving route attribute determination unit 42 determines whether the value of t 2 (that is, the difference until the time when the next record is recorded) included in the movement vector function value of the record is larger than the effective record threshold time T. If the value of t 2 (that is, the difference until the time when the next record is recorded) included in the movement vector function value of the record is larger than the effective record threshold time T, the staying / moving route attribute determination unit 42 It is impossible to judge the attribute of. On the other hand, if the t 2 value included in the movement vector function value of the record is smaller than the effective record threshold time T, the residence / movement route attribute determination unit 42 further determines the r value and the residence point included in the movement vector function value. The value of the allowable distance R is compared.
- the staying / moving route attribute determination unit 42 determines that the target record is a record having the staying attribute.
- the staying / moving route attribute determination unit 42 determines that the target record is a record having the moving route attribute. Then, the staying / moving route attribute determination unit 42 inputs a record determined to be the staying attribute or the moving attribute to the clustering space plotting unit 21.
- the action pattern analysis unit 50 includes a clustering space plot unit 21, a retention point cluster extraction unit 22, and a movement route cluster extraction unit 24. That is, the behavior pattern analysis unit 50 in the present embodiment is different from the behavior pattern analysis unit 20 in the first embodiment in that the retention point position information log removal unit 23 is not included.
- the clustering space plotting unit 21 plots the position information log to be analyzed in the clustering space. Specifically, first, the clustering space plotting unit 21 sets a preset time (that is, time reference) as a date break, and extracts a position information log of the set day (that is, day reference). The method of extracting the position information log based on the time reference and the day of the week reference is the same as the method described in the first embodiment.
- the clustering space plotting unit 21 may directly receive the position information log from the position information log input unit 10. Alternatively, the staying / moving route attribute determining unit 42 may input the position information log input to the moving vector operation analysis unit 40 to the clustering space plotting unit 21.
- the clustering space plotting unit 21 plots, in the clustering space, a record in which the traveling direction angle received from the staying / moving route attribute determining unit 42 is added to the position information log including latitude, longitude, and time. .
- the traveling direction angle is a value corresponding to the time included in the position information log. That is, the clustering space plotting unit 21 plots a record including [latitude, longitude, time, traveling direction angle] in a four-dimensional clustering space. Note that the method by which the clustering space plotting unit 21 plots the records in the clustering space is the same as that in the first embodiment. Then, the clustering space plotting unit 21 inputs the plotted data to the staying point cluster extracting unit 22.
- the retention point cluster extraction unit 22 performs clustering of clustering spatial data as in the first embodiment, and extracts retention points. However, in the present embodiment, when the retention point cluster extraction unit 22 receives the data plotted in the clustering space from the clustering space plotting unit 21, only the records determined as the retention attribute by the retention / movement route attribute determination unit 42 are Perform clustering as a target.
- the staying point cluster extraction unit 22 multiplies each value indicating the latitude and longitude of the data plotted in the clustering space by a preset weight value.
- data determined as the stagnation attribute is set as an initial cluster.
- the staying point cluster extraction unit 22 determines whether the Euclidean distance between two points selected from the clusters does not exceed a preset Euclidean distance threshold. If the Euclidean distance does not exceed the Euclidean distance threshold, the staying point cluster extraction unit 22 integrates clusters having the shortest Euclidean distance between two points selected from all clusters into one cluster.
- the retention point cluster extraction unit 22 integrates the clusters so as to include the time, latitude, and longitude of each position information log that configures the clusters. Furthermore, the staying point cluster extraction unit 22 calculates the value of the gravity center of latitude and longitude from the position information log included in the cluster separately from each position information log, and uses the value as a new cluster data value. . The retention point cluster extraction unit 22 also holds the number of pieces of clustering spatial data included in the integrated cluster.
- the staying point cluster extraction unit 22 determines that a cluster including clustering spatial data whose number exceeds the set cluster number threshold is a staying point. Then, the retention point cluster extraction unit 22 sets the earliest time among the data in the cluster as the retention start time, and the latest time as the retention end time. Then, the retention point cluster extraction unit 22 causes the retention point pattern storage unit 31 to store the retention start time, the retention end time, and the values of the centers of gravity of latitude and longitude in the data in the cluster.
- the staying point cluster extraction unit 22 inputs the data received from the clustering space plotting unit 21 to the movement path cluster extraction unit 24.
- the method by which the staying point cluster extraction unit 22 performs clustering in the present embodiment is the same as that in the first embodiment.
- the present embodiment is different from the first embodiment in that clustering is performed only on records determined to be the retention attribute by the retention point cluster extraction unit 22.
- the movement path cluster extraction unit 24 performs clustering of clustering spatial data as in the first embodiment, and extracts a time zone and movement path in which the terminal user is moving frequently. However, it differs from the first embodiment in that data determined as a movement attribute is processed as an initial cluster.
- the retention point position information log removing unit 23 subtracts the data determined to be the retention point included in the retention point pattern data from the data plotted in the clustering space. Then, the staying point position information log removing unit 23 inputs the remaining data subtracted from the data plotted in the clustering space to the moving route cluster extracting unit 24.
- a record in which the attribute of residence or movement route is determined by the residence / movement route attribute determination unit 42 is input to the clustering space plotting unit 21. Therefore, even if the data determined to be the retention point by the retention point position information log removal unit 23 is not subtracted from the data plotted in the clustering space, the movement path cluster extraction unit 24 identifies the data determined as the movement attribute It will be possible to
- the staying point position information log removing unit 23 in the first embodiment is also the moving vector computing unit 40 in the present embodiment (more specifically, the moving vector function value computing unit 41 and the staying / moving route attribute determining unit 42) Also has the function of extracting non-dwelling points. Furthermore, in the movement vector computing unit 40 in the present embodiment, since data is narrowed down, in addition to the effects in the first embodiment, even if the position information log becomes large, The determination accuracy can be improved.
- the movement route cluster extraction unit 24 multiplies the latitude and longitude of the received data by the preset weight value.
- the traveling direction of the moving route is considered. Therefore, the distance function is defined as Equation 1 below using four variables: latitude, longitude, time, and travel direction angle.
- k 2 and k 3 are respectively determined for the time and the traveling direction angle with reference to the latitude and the longitude.
- the movement path cluster extraction unit 24 determines whether the Euclidean distance between two points selected from the clusters does not exceed the set Euclidean distance threshold. When the Euclidean distance does not exceed the threshold value, the movement path cluster extraction unit 24 integrates clusters having the shortest Euclidean distance between two points selected from among the clusters into one cluster.
- the movement path cluster extraction unit 24 integrates the clusters so as to include the time, latitude, and longitude of each position information log that configures the clusters. Further, the movement route cluster extraction unit 24 calculates the value of the gravity center of latitude and longitude from the position information log included in the cluster separately from each position information log, and uses the value as a new cluster data value. . In addition, the movement path cluster extraction unit 24 also holds the number of clustering spatial data included in the integrated cluster.
- the movement path cluster extraction unit 24 determines that a cluster including the clustering spatial data whose number exceeds the set cluster number threshold is a movement path. Then, the movement route cluster extraction unit 24 sets the earliest time of the data in the cluster as the movement start time, and the latest time as the movement end time.
- the movement path cluster extraction unit 24 may be configured to include position information (latitude and longitude) of the cluster and position information (latitude and longitude) of the staying point included in the staying point pattern data received from the staying point cluster extracting unit 22. Compare. Then, the movement path cluster extraction unit 24 determines, as the first adjacent staying point identifier and the second adjacent staying point identifier, the staying point having the first Euclidean distance with the position of the cluster and the second shortest staying point, respectively. . Then, the movement path cluster extraction unit 24 causes the movement path pattern storage unit 32 to store the movement start time, the movement end time, the first proximity dwell point identifier, and the second proximity dwell point identifier.
- the present embodiment differs from the movement route cluster extraction unit 24 in the first embodiment in that the movement route cluster extraction unit 24 performs clustering in consideration of the difference in the traveling direction angle.
- the moving route cluster extraction unit 24 may perform clustering based on the positioning position of the user and the positioning date and time, as in the first embodiment.
- the movement vector function value calculation unit 41 is sorted in the order of positioning date and time included in the position information log.
- the time difference between the positioning dates and times included in the position information log, the distance between the positioning locations, and the traveling direction angle representing the traveling direction between the positioning locations are calculated.
- the staying / moving route attribute determination unit 42 determines whether the attribute of the position information log has the staying attribute, based on the time difference between the positioning dates and times and the distance between the positioning locations.
- the retention point cluster extraction unit 22 extracts a retention point by clustering the position information logs determined to have the retention attribute.
- the staying / moving route attribute determining unit 42 determines whether the attribute of the position information log has the moving attribute, based on the time difference between the positioning dates and times and the distance between the positioning locations.
- the clustering space plotting unit 21 defines a record obtained by adding the traveling direction angle to the position information log including the user's positioning position and positioning date and time. Plot to Then, the movement route cluster extraction unit 24 extracts the movement route of the user by clustering the position information logs determined to have the movement route attribute.
- the determination accuracy of the staying point cluster and the movement route cluster can be improved even when the position information log is large.
- the record which can not be determined which record indicates the retention or the movement path (that is, the record having neither the retention attribute nor the movement attribute) is excluded from the clustering target. In this way, by reducing the records targeted for clustering, the accuracy of clustering can be improved.
- FIG. 12 is an explanatory diagram of an example of aggregation of position information logs.
- dotted arrows indicate the traveling directions included in the position information log.
- the position information log 401a included in the position information log group 401 surrounded by the broken line and the position information log 401b have the same traveling direction but use different routes. Indicates Further, in the example shown in FIG.
- the position information log 402a included in the position information log group 402 surrounded by the broken line and the position information log 402b indicate that the traveling direction is largely changed on the same movement route.
- another route is classified as another movement route cluster included in the position information log group 401, and the process of improving the accuracy of classifying the same movement route included in the position information log group 401 as the same movement route cluster Will be explained.
- FIG. 13 is a block diagram showing a configuration example of a third embodiment of the behavior pattern analysis device according to the present invention.
- the behavior pattern analysis device in the present embodiment includes a position information log input unit 10, a behavior pattern analysis reference input unit 11, a behavior pattern analysis unit 60, a behavior pattern storage unit 30, and a movement vector calculation analysis unit 40. ing.
- Movement vector calculation analysis unit 40 including position information log input unit 10, action pattern analysis standard input unit 11, action pattern storage unit 30, movement vector function value calculation unit 41 and stagnation / movement route attribute determination unit 42 Is the same as that of the second embodiment, and thus the description thereof is omitted.
- the action pattern analysis unit 60 includes a clustering space plot unit 21, a staying point cluster extraction unit 22, a movement route cluster extraction unit 24, and a traveling direction distance coefficient calculation unit 25. That is, the behavior pattern analysis unit 60 in the present embodiment is different from the behavior pattern analysis unit 50 in the second embodiment in that the movement direction distance coefficient calculation unit 25 is included.
- the clustering space plotting unit 21 plots the position information log to be analyzed in the clustering space, as in the second embodiment. Specifically, first, the clustering space plotting unit 21 sets a preset time (that is, time reference) as a date break, and extracts a position information log of the set day (that is, day reference). The method of extracting the position information log based on the time reference and the day of the week reference is the same as the method described in the second embodiment.
- the clustering space plotting unit 21 may directly receive the position information log from the position information log input unit 10. Alternatively, the staying / moving route attribute determining unit 42 may input the position information log input to the moving vector operation analysis unit 40 to the clustering space plotting unit 21.
- the clustering space plotting unit 21 plots position information logs including latitude, longitude and time in a clustering space.
- the method of the clustering space plotting unit 21 plotting the above records in the clustering space is the same as that of the first embodiment. Then, the clustering space plotting unit 21 inputs the plotted data to the staying point cluster extracting unit 22.
- the retention point cluster extraction unit 22 performs clustering of clustering spatial data as in the second embodiment to extract retention points.
- the retention point cluster extraction unit 22 receives data plotted in the clustering space from the clustering space plotting unit 21, only the records determined as the retention attribute by the retention / movement route attribute determination unit 42 are targeted. Perform clustering.
- the staying point cluster extraction unit 22 multiplies each value indicating the latitude and longitude of the data plotted in the clustering space by a preset weight value.
- data determined as the stagnation attribute is set as an initial cluster.
- the staying point cluster extraction unit 22 determines whether the Euclidean distance between two points selected from the clusters does not exceed a preset Euclidean distance threshold. If the Euclidean distance does not exceed the Euclidean distance threshold, the staying point cluster extraction unit 22 integrates clusters having the shortest Euclidean distance between two points selected from all clusters into one cluster.
- the retention point cluster extraction unit 22 integrates the clusters so as to include the time, latitude, and longitude of each position information log that configures the clusters. Furthermore, the staying point cluster extraction unit 22 calculates the value of the gravity center of latitude and longitude from the position information log included in the cluster separately from each position information log, and uses the value as a new cluster data value. . The retention point cluster extraction unit 22 also holds the number of pieces of clustering spatial data included in the integrated cluster.
- the staying point cluster extraction unit 22 determines that a cluster including clustering spatial data whose number exceeds the set cluster number threshold is a staying point. Then, the retention point cluster extraction unit 22 sets the earliest time among the data in the cluster as the retention start time, and the latest time as the retention end time. Then, the retention point cluster extraction unit 22 causes the retention point pattern storage unit 31 to store the retention start time, the retention end time, and the values of the centers of gravity of latitude and longitude in the data in the cluster.
- the staying point cluster extraction unit 22 inputs the data received from the clustering space plotting unit 21 to the movement path cluster extraction unit 24.
- the method by which the staying point cluster extraction unit 22 performs clustering in the present embodiment is the same as that in the second embodiment.
- the clustering space plotting unit 21 plots the record including [latitude, longitude, time, traveling direction angle] in the four-dimensional clustering space
- the clustering space plotting unit 21 differs from the second embodiment in that the plotting unit 21 performs clustering by plotting records including [latitude, longitude, time] in a three-dimensional clustering space.
- the movement path cluster extraction unit 24 performs clustering of clustering spatial data, and extracts a time zone and movement path in which the terminal user is moving frequently.
- the present embodiment differs from the second embodiment in that the movement route cluster extraction unit 24 performs clustering in consideration of the relationship between the traveling direction indicated by the position information log and the azimuth angle between the position information logs. .
- FIG. 14 is an explanatory view showing an example of the relationship between the traveling direction indicated by the position information log and the azimuth angle of the position information log.
- arrows 501a and 501b indicate the traveling directions of the points P1 and P2 indicating the position information log.
- the traveling direction distance coefficient calculation unit 25 calculates the traveling direction distance coefficient w 12 from the traveling direction angle ⁇ 1 of P 1 and the azimuth angle ⁇ 12 to P 2 having P 1 as the origin, and moves The path cluster extraction unit 24 multiplies w 12 by the value of the Euclidean distance described above. Similarly, the traveling direction distance coefficient calculation unit 25 calculates the traveling direction distance coefficient w 21 from the traveling direction angle ⁇ 2 of P 2 and the azimuth angle ⁇ 21 to P 2 with P 1 as the origin, and the movement path cluster extraction unit 24 multiplies w 21 by the aforementioned Euclidean distance value.
- the traveling direction distance coefficient is a coefficient whose value is smaller as the value of the angle formed by the traveling direction from the positioning position indicated by the position information log and the direction connecting the two positioning positions is smaller, which will be described later. Calculated by the method. Then, using the calculated value, the movement route cluster extraction unit 24 performs processing in which the data determined as the movement attribute is set as the initial cluster. The method of calculating the azimuth will also be described later.
- the traveling direction distance coefficient calculation unit 25 determines the angle between the traveling direction from the positioning position indicated by each position information log and the direction connecting the positioning positions indicated by the two position information logs.
- a coefficient that is, a traveling direction distance coefficient
- the traveling direction distance coefficient calculation unit 25 sets the vector to another point whose origin is one point to the position information log of two points (specifically, the direction between two positioning positions is Calculation of the angle formed by the moving vector function value calculating unit 41 and the moving direction vector (specifically, the vector indicating the moving direction from the positioning position), the value of the moving direction distance coefficient Decide.
- the traveling direction distance coefficient calculation unit 25 receives the position information log from the moving route cluster extraction unit 24 and calculates the traveling direction distance coefficient.
- the movement direction vectors of the two position information logs are usually different. Therefore, the traveling direction distance coefficient calculation unit 25 performs calculation of the angle to be formed, with each position information log as an origin.
- the vector to P2 with the origin of the P1 is defined by the formula 3 below.
- the traveling direction distance coefficient calculation unit 25 uses the theta 1to2 calculated by Equation 2 and Equation 3, obtains the traveling direction distance factor w 12.
- the formula used to calculate the traveling direction coefficients w 12 is not limited to Equation 4 above.
- any other equation may be used as long as the value of the traveling direction distance coefficient always takes a positive value regardless of the angle between the two.
- the traveling direction distance coefficient is formed by the traveling direction included in the movement vector function value calculated by the movement vector function value calculator 41 and the azimuth angle of the two position information logs. It can be said that the coefficient is determined from the value of the corner.
- Equation 5 the azimuth angle of P1 with P2 as the origin is defined as Equation 5 below.
- ⁇ 21 arg ((Y 1 -Y 2 ) / (X 1 -X 2 )) (Equation 5)
- angle ⁇ 2 to 1 between the vector to P 1 having the origin at P 2 and the movement direction vector 501 b of P 2 calculated by the movement vector function value calculator 41 is also defined by the following expression 6 as in expression 3. Be done.
- ⁇ 2 to 1 min ⁇
- the traveling direction distance coefficient calculation unit 25 obtains the traveling direction distance coefficient w 21 using ⁇ 2 to 1 calculated by Equation 5 and Equation 6.
- the traveling direction distance coefficient calculation unit 25 may calculate the traveling direction distance coefficient w 21 using, for example, the following Expression 7.
- the traveling direction distance coefficient calculation unit 25 notifies the travel route cluster extraction unit 24 of the calculated traveling direction distance coefficients w 12 and w 21 .
- the movement route cluster extraction unit 24 Upon receiving the data from the staying point cluster extraction unit 22, the movement route cluster extraction unit 24 passes the position information log including at least [latitude, longitude, movement direction angle] to the movement direction distance coefficient calculation unit 25, and the movement direction The traveling direction distance coefficient is received from the distance coefficient calculating unit 25.
- the calculation when calculating a value indicating the distance between two points (P1, P2) of the position information log, the case where the distance is calculated based on P1 and the case where the distance is calculated based on P2 When considering the traveling direction of the movement path, the calculation is performed so that the value indicating the distance between the two is different.
- a distance function is defined like the following formula 8 using four variables of latitude, longitude, time, and a traveling direction distance coefficient.
- Equation 8 based on the latitude and longitude, relative to time, k 2 are respectively determined.
- movement route cluster extraction unit 24 by multiplying the weighting value k 2 respectively the latitude and longitude included in the position information log, calculates the Euclidean distance between the position information log. Furthermore, the movement route cluster extraction unit 24 performs weighting by multiplying the calculated Euclidean distance by the traveling direction distance coefficient.
- the movement route cluster extraction unit 24 determines whether or not the Euclidean distance between two points selected from the above clusters exceeds the set Euclidean distance threshold. When the Euclidean distance does not exceed the threshold value, the movement path cluster extraction unit 24 integrates clusters having the shortest Euclidean distance between two points selected from among the clusters into one cluster.
- the movement path cluster extraction unit 24 integrates the clusters so as to include the time, latitude, and longitude of each position information log that configures the clusters. Further, the movement route cluster extraction unit 24 calculates the value of the gravity center of latitude and longitude from the position information log included in the cluster separately from each position information log, and uses the value as a new cluster data value. . In addition, the movement path cluster extraction unit 24 also holds the number of clustering spatial data included in the integrated cluster.
- the movement path cluster extraction unit 24 determines that a cluster including the clustering spatial data whose number exceeds the set cluster number threshold is a movement path. Then, the movement route cluster extraction unit 24 sets the earliest time of the data in the cluster as the movement start time, and the latest time as the movement end time.
- the movement path cluster extraction unit 24 may be configured to include position information (latitude and longitude) of the cluster and position information (latitude and longitude) of the staying point included in the staying point pattern data received from the staying point cluster extracting unit 22. Compare. Then, the movement path cluster extraction unit 24 determines, as the first adjacent staying point identifier and the second adjacent staying point identifier, the staying point having the first Euclidean distance with the position of the cluster and the second shortest staying point, respectively. . Then, the movement path cluster extraction unit 24 causes the movement path pattern storage unit 32 to store the movement start time, the movement end time, the first proximity dwell point identifier, and the second proximity dwell point identifier.
- the movement route cluster extraction unit 24 of the present embodiment performs Euclidean distance calculation on records including [latitude, longitude, time], and performs clustering of these, the movement route in the second embodiment. It differs from the cluster extraction unit 24.
- the movement route cluster extraction unit 24 according to the present embodiment calculates the Euclidean distance between two position information logs using each position information log as a start point, and further multiplying this value by the traveling direction distance coefficient. Define the distance between two points.
- the second embodiment differs from the movement route cluster extraction unit 24 in the second embodiment in that clustering is performed in consideration of two values defined for each point. The other processes are similar to those of the second embodiment.
- the movement direction vectors do not point in the direction of each other position information log. Therefore, the Euclidean distance (see Equation 2 above) multiplied by the traveling direction distance coefficient is far regardless of which of the position information log 401a and the position information log 401b is used as the traveling direction distance coefficient. Therefore, position information logs of two points are less likely to be clustered.
- the movement direction vector of the position information log 402a is directed to the position information log 402b.
- the movement direction vector of the position information log 402b is directed away from the direction of the position information log 402a. Therefore, although the value of the traveling direction distance coefficient calculated based on the position information log 402b increases, the value of the traveling direction distance coefficient calculated based on the position information log 402a decreases.
- the movement path cluster extraction unit 24 performs clustering in the order of small values of Euclidean distance multiplied by the traveling direction distance coefficient, and as a result, positional information logs of two points are easily clustered.
- the traveling direction distance coefficient calculation unit 25 calculates the traveling direction distance coefficient for each position information log based on the two position information logs.
- the movement route cluster extraction unit 24 performs weighting on the position information log that makes it easy to determine that the Euclidean distance in the time direction is far based on the latitude and longitude, and the Euclidean distance coefficient to the Euclidean distance Perform weighting by multiplication. Then, the movement route cluster extraction unit 24 extracts the movement route of the user by clustering the position information logs determined to have the non-staying attribute among the weighted position information logs. Therefore, in addition to the effect of the first embodiment, it is possible to improve the accuracy of classifying the movement route cluster.
- the position information log input unit 10 inputs the position information log to the clustering space plotting unit 21.
- the clustering space plotting unit 21 takes a preset time as a date break, and extracts a position information log of the set day of the week. Then, the clustering space plotting unit 21 plots a record including [latitude, longitude, time] in a three-dimensional clustering space, and inputs the plotted data to the staying point cluster extracting unit 22.
- values indicating latitude and longitude of the data plotted in the clustering space are respectively multiplied by preset weight values.
- all data plotted in the clustering space and multiplied by the set weight value is set as an initial cluster.
- the staying point cluster extraction unit 22 determines whether the Euclidean distance between two points selected from all clusters does not exceed the Euclidean distance threshold set in advance. When the Euclidean distance does not exceed the threshold value, the staying point cluster extraction unit 22 integrates clusters having the shortest Euclidean distance between two points selected from among all clusters into one cluster.
- the retention point cluster extraction unit 22 integrates the clusters so as to include the time, latitude, and longitude of each position information log that configures the clusters. Furthermore, the staying point cluster extraction unit 22 calculates the value of the gravity center of latitude and longitude from the position information log included in the cluster separately from each position information log, and uses the value as a new cluster data value. . The retention point cluster extraction unit 22 also holds the number of pieces of clustering spatial data included in the integrated cluster.
- the retention point cluster extraction unit 22 determines a cluster including clustering spatial data whose number exceeds the set cluster number threshold as the retention point. Then, the retention point cluster extraction unit 22 sets the earliest time among the data in the cluster as the retention start time, and the latest time as the retention end time. Then, the retention point cluster extraction unit 22 causes the retention point pattern storage unit 31 to store the retention start time, the retention end time, and the values of the centers of gravity of latitude and longitude in the data in the cluster.
- the retention point cluster extraction unit 22 retains the data received from the clustering space plotting unit 21 (that is, the data plotted in the clustering space) and the data determined to be a retention point (retention point pattern data) from the retention point position information log removal unit Enter 23
- the retention point position information log removing unit 23 subtracts the data determined to be the retention point included in the retention point pattern data from the data plotted in the clustering space. Then, the staying point position information log removing unit 23 inputs the remaining data subtracted from the data plotted in the clustering space to the movement route cluster extracting unit 24.
- the movement route cluster extraction unit 24 receives data from the staying point position information log removal unit 23
- the movement route cluster extraction unit 24 multiplies the latitude and longitude of the received data by the preset weight value.
- all data to which the weight value has been multiplied is taken as an initial cluster.
- the movement path cluster extraction unit 24 determines whether the Euclidean distance between two points selected out of all the clusters exceeds the set Euclidean distance threshold. When the Euclidean distance does not exceed the threshold value, the movement path cluster extraction unit 24 integrates clusters having the shortest Euclidean distance between two points selected from among all clusters into one cluster. At this time, the movement path cluster extraction unit 24 integrates the clusters so as to include the time, latitude, and longitude of each position information log that configures the clusters. Further, the movement route cluster extraction unit 24 calculates the value of the gravity center of latitude and longitude from the position information log included in the cluster separately from each position information log, and uses the value as a new cluster data value. . In addition, the movement path cluster extraction unit 24 also holds the number of clustering spatial data included in the integrated cluster.
- the movement route cluster extraction unit 24 determines a cluster including the clustering spatial data of the number exceeding the set cluster number threshold as the movement route. Then, the movement route cluster extraction unit 24 sets the earliest time of the data in the cluster as the movement start time, and the latest time as the movement end time.
- the movement route cluster extraction unit 24 receives position information (latitude and longitude) of the cluster and position information (latitude and longitude) of the staying point included in the staying point pattern data received from the staying point position information log removing unit 23. Compare with. Then, the movement path cluster extraction unit 24 moves the moving path by the first short staying point and the second shortest staying point with the Euclidean distance to the position of the cluster, the first near staying point identifier and the second near staying point identifier, respectively. Decide. Then, the movement path cluster extraction unit 24 causes the movement path pattern storage unit 32 to store the movement start time, the movement end time, the first proximity dwell point identifier, and the second proximity dwell point identifier.
- FIG. 15 is a block diagram showing an example of the minimum configuration of the behavior pattern analysis device according to the present invention.
- the behavioral pattern analysis device according to the present invention is a numerical value information (for example, a position information log representing a position information log which is information including a user's positioning position (for example, latitude, longitude, altitude) and positioning date and time (for example, time).
- Position information plotting means 81 (eg, clustering space plotting unit 21) plotting on a multidimensional space (eg, three-dimensional space defined by latitude, longitude, and time) defined by The Euclidean distance in the time direction (for example, the time axis direction) with respect to the position information space (for example, the latitude-longitude plane) which is a space defined by the numerical information representing the positioning position in the multidimensional space Weighting (for example, processing of multiplying time by weight value k) is performed on the position information log, and the weighted position information log is A dwelling point cluster extracting unit 82 (for example, a dwelling point cluster extracting unit 22) which extracts a dwelling point which is a position where the user frequently dwells by tumbling, and the position information log plotted by the position information plotting unit 81
- Non-dwelling point position information log extracting means 83 (for example, staying point position information log removing unit 23) for extracting a set of position information logs excluding position information logs extracted as staying points
- FIG. 16 is a block diagram showing an example of another minimum configuration of the behavior pattern analysis device according to the present invention.
- Another behavior pattern analysis device according to the present invention is sorted in the order of positioning date and time included in the position information log, which is information including the user's positioning position (for example, latitude, longitude, altitude) and positioning date and time (for example, time).
- movement vector function value calculation means 91 for example, movement vector function value for calculating the time difference between positioning dates and times included in the position information log and the distance between positioning positions Whether the attribute of the position information log indicates the staying point indicating the staying point which is the position at which the user frequently stays or the moving route of the user, based on the operation unit 41), the time difference between the positioning date and time, and the distance between the positioning positions
- Attribute determination means 92 for example, residence / movement route attribute determination unit 42
- positioning position Position information plotting means 93 eg, plotting on a multi-dimensional space (eg, three-dimensional space defined by latitude, longitude and time) defined by numerical information (eg, latitude, longitude, altitude) representing time The Euclidean distance in the time direction (for example, the time axis direction)
- the movement vector function value calculation unit 91 may calculate, as a difference value between two position information logs, a difference value of a traveling direction angle (for example, ⁇ ) representing a traveling direction between positioning positions. Then, the position information plotting means 93 adds a traveling direction angle to the position information log including the user's positioning position and positioning date and time (for example, a record including [latitude, longitude, time, traveling direction angle]), It may be plotted on a multidimensional space (eg, a four-dimensional space defined by latitude, longitude, altitude, and time) defined by numerical information representing a positioning position, time, and a traveling direction angle. Furthermore, the movement route cluster extraction unit 95 may extract the movement route of the user by clustering records obtained by adding the traveling direction angle to the position information log determined to have the movement route attribute.
- a traveling direction angle for example, ⁇
- a position information plotting means for plotting a position information log which is information including a user's positioning position and positioning date and time, in a multidimensional space defined by numerical information representing the positioning position and time.
- a weighting is performed on the position information log to make it easy to determine that the Euclidean distance in the time direction is close to the position information space which is a space defined by the numerical information representing the positioning position in a two-dimensional space.
- a retention point cluster extraction means for extracting a retention point which is a position where the user frequently stays by clustering the position information log extracted as the retention point from the position information log plotted by the position information plotting means
- a non-dwelling point positional information log extracting means for extracting a set of excluded positional information logs as a non-dwelling point; The moving path of the user is extracted by performing weighting on the position information log of the non-dwelling point, which makes it easy to determine that the Euclidean distance in the time direction to the information space is long, and clustering the weighted position information log.
- a movement pattern analysis device characterized by comprising movement route cluster extraction means.
- the positional information plotting means plots a positional information log including values indicating latitude and longitude as a user's positioning position on a three-dimensional space defined by latitude, longitude and time, and extracts a stay point cluster
- Analysis device is used to analyze the position information log.
- the positional information plotting means plots a positional information log including values indicating latitude, longitude and altitude as a user's positioning position on a four-dimensional space defined by latitude, longitude, altitude and time
- the retention point cluster extraction means adds weight to the position information log which makes it easy to determine that the Euclidean distance in the time direction is close to the latitude, longitude, and altitude space which is a space defined by the latitude, longitude, and altitude in the four-dimensional space.
- the action pattern analyzer according to appendix 1, which is performed.
- the movement route cluster extraction unit determines the retention point having the shortest Euclidean distance with the clustered position information log and the second shortest retention point as the start point or the end point of the movement route.
- the action pattern analyzer according to any one of the four.
- the movement route cluster extraction means compares the value of the center of gravity of the position information in the clustered position information log with the distance of the value of the center of gravity of the position information at the retention point,
- the behavior pattern analyzer according to any one of appendices 1 to 5, wherein the point and the second shortest dwell point are determined as the start point or the end point of the movement path.
- a difference value between two adjacent position information logs sorted in the order of the positioning date and time included in the position information log that is information including the user's positioning position and positioning date and time is included in the position information log
- the attribute of the position information log is a user based on the time difference between the positioning dates and times, the movement vector function value calculating means for calculating the distance between the positioning positions, and the time difference between the positioning dates and times and the distance between the positioning positions
- Attribute determination means for determining whether it is a staying attribute indicating a staying point at which the user frequently dwells or a non-dwelling attribute indicating a non-staying point indicating a position on the moving path of the user
- Position information plotting means for plotting in a multidimensional space defined by numerical information representing a positioning position and time, and defined by numerical information representing the positioning position in the multidimensional space
- a position information log which is determined to have the retention attribute among the weighted position information logs by performing weighting on the position information log which makes it easy to determine that the
- the movement pattern cluster extraction means for extracting the movement path of the user by clustering the position information logs determined to have the non-staying attribute among the position information logs .
- the movement vector function value calculation means calculates, as a difference value between two position information logs, a difference value of an advancing direction angle representing an advancing direction between positioning positions, and the position information plotting means A record obtained by adding the traveling direction angle to the position information log including the positioning position and the positioning date and time is plotted on a multidimensional space defined by numerical information representing the positioning position, time, and the traveling direction angle.
- the activity pattern analysis device according to appendix 7, wherein the means extracts the moving route of the user by clustering records obtained by adding the traveling direction angle to the position information log determined to have the moving route attribute.
- the traveling direction distance coefficient calculating means is provided for calculating, for each position information log, an advancing direction distance coefficient which is a coefficient whose value decreases, and the movement route cluster extracting means determines that the Euclidean distance in the time direction to the position information space is far Weighting is performed on the position information log so as to be easily performed, and weighting is performed by multiplying the Euclidean distance by the traveling direction distance coefficient calculated by the moving direction distance coefficient calculating means, and weighting of the weighted position information log Addendum 7 of extracting user's moving route by clustering position information logs determined to have non-dwelling attribute among them. Of behavior pattern analysis apparatus.
- a position information log which is information including a user's positioning position and positioning date and time, is plotted on a multidimensional space defined by numerical information representing the positioning position and time, and the positioning in the multidimensional space
- weighting on the position information log, which makes it easy to determine that the Euclidean distance in the time direction is close to the position information space which is a space defined by numerical information representing position
- clustering the weighted position information logs Extracting a retention point which is a position at which the user frequently dwells, and removing a set of position information logs excluding the position information log extracted as the retention point from the position information log plotted on the multidimensional space.
- a position information log including values indicating latitude and longitude as the user's positioning position is plotted on a three-dimensional space defined by latitude, longitude and time, and defined by the latitude and longitude in the three-dimensional space 10.
- a position information log including values indicating latitude, longitude and altitude as the user's positioning position is plotted on a four-dimensional space defined by the latitude, longitude, altitude and time, and the latitude in the four-dimensional space 10.
- a difference value between two adjacent position information logs sorted in the order of the positioning date and time included in the position information log that is information including the user's positioning position and positioning date and time is included in the position information log
- the time difference between the positioning dates and times, and the distance between the positioning positions are calculated, and based on the time difference between the positioning dates and times and the distance between the positioning positions It is determined whether it is a staying attribute indicating a certain staying point or a non-dwelling attribute indicating a non-dwelling point indicating a position on the moving route of the user, and the position information log is defined by numerical information representing the positioning position and time.
- the Euclidean distance in the time direction to the position information space which is a space defined by the numerical information representing the positioning position in the multidimensional space
- the retention point is extracted by performing weighting with respect to the position information log to facilitate the process and clustering the position information log determined to have the retention attribute among the weighted position information logs, thereby extracting the retention point, and the position information Performing weighting on the position information log to easily determine that the Euclidean distance in the time direction with respect to space is far, and clustering the position information log determined to have the non-dwelling attribute among the weighted position information logs
- An action pattern analysis method characterized by extracting a movement route of a user by
- the position information log is weighted so that it is easily determined that the Euclidean distance in the time direction is close to the position information space which is a space defined by numerical information representing the positioning position in the multidimensional space.
- a retention point cluster extraction process for extracting a retention point which is a position at which a user frequently stays by clustering information logs, and position information extracted as the retention point from a position information log plotted by position information plot processing
- Non-dwelling point location information log extraction that extracts a set of location information logs excluding logs as non-dwelling points
- a user is weighted by performing processing and weighting that makes it easy to determine that the Euclidean distance in the time direction with respect to the position information space is far from the position information log of the non-staying point, and clustering the weighted position information log
- route cluster extraction process which extracts the movement path
- position information logs including values indicating latitude and longitude as the user's positioning position are plotted on a three-dimensional space defined by latitude, longitude, and time
- the point cluster extraction process performs weighting on the position information log that makes it easy to determine that the Euclidean distance in the time direction with respect to the latitude and longitude plane which is a plane defined by the latitude and longitude in the three-dimensional space is short. Behavior pattern analysis program described.
- a location information log including values indicating latitude, longitude, and altitude as a user's positioning location in location information plotting processing.
- the position information log is weighted to make it easy to determine that Euclidean distance in the time direction is close to the latitude, longitude, and altitude space, which is a space defined by the latitude, longitude, and altitude in the four-dimensional space.
- the positional information log is a difference value between two adjacent positional information logs sorted in the order of the positioning date and time included in the positional information log that is information including the user's positioning position and positioning date and time on the computer.
- the attribute of the position information log is based on the time difference of the positioning date and time included in the movement vector, the movement vector function value calculation processing for calculating the distance between the positioning positions, the time difference of the positioning date and time, and the distance between positioning positions; Attribute determination processing for determining whether a residence attribute indicates a residence point at which a user frequently resides or a non-dwelling attribute indicates a non-dwelling point at a position on the user's movement route; Position information plotting processing for plotting on the multidimensional space defined by the numerical information representing the positioning position and the time, the numerical information representing the positioning position in the multidimensional space It is determined that the position information log is weighted so that the Euclidean distance in the time direction with respect to the position information space which is a defined space is likely to be close, and that the position attribute log has the retention attribute among the weighted position information logs.
- a retention point cluster extraction process for extracting the retention point by performing clustering on the position information log, and weighting for making it easy to determine that the Euclidean distance in the time direction with respect to the position information space is long are performed on the position information log
- the difference value of the traveling direction angle representing the advancing direction between the positioning positions is calculated as the difference value between the two position information logs, and the position information plotting process is performed.
- the traveling direction distance coefficient calculation processing is performed to calculate, for each position information log, a traveling direction distance coefficient, which is a coefficient whose value decreases as the distance is smaller, and Euclidean distance in the time direction to the position information space in the movement route cluster extraction processing. Is given to the position information log, and weighting is performed by multiplying the Euclidean distance by the traveling direction distance coefficient calculated in the movement direction distance coefficient calculation processing, and the weighting is performed.
- Behavioral pattern analysis program note 21 wherein for the extraction path.
- the present invention is suitably applied to an action pattern analysis device that analyzes an action pattern from position information measured irregularly. For example, even when the positioning interval of the position information using the terminal with the position information acquisition function is wide or when the position information is irregularly measured, it is possible to grasp the user's action pattern. Therefore, the accuracy of the recommendation service for the purpose of content distribution matching the user's behavior pattern is improved.
Landscapes
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Radar, Positioning & Navigation (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Navigation (AREA)
- Traffic Control Systems (AREA)
Abstract
Description
図1は、本発明による行動パタン解析装置の第1の実施形態の構成例を示すブロック図である。本実施形態における行動パタン解析装置は、位置情報ログ入力部10と、行動パタン解析基準入力部11と、行動パタン解析部20と、行動パタン記憶部30とを備えている。
次に、本発明による行動パタン解析装置の第2の実施形態を説明する。本実施形態では、位置情報ログが大量に存在する場合に、滞留点クラスタと移動経路クラスタの判別精度を上げる手法を説明する。図10は、本発明による行動パタン解析装置の第2の実施形態の構成例を示すブロック図である。本実施形態における行動パタン解析装置は、位置情報ログ入力部10と、行動パタン解析基準入力部11と、行動パタン解析部50と、行動パタン記憶部30と、移動ベクトル演算解析部40とを備えている。
但し|θ1-θ2|<180°
(X1-X2)2+(Y1-Y2)2+k2(Z1-Z2)2+k3(|θ1-θ2|-180)2
但し|θ1-θ2|>180° ・・・(式1)
次に、本発明による行動パタン解析装置の第3の実施形態を説明する。本実施形態では、各移動経路クラスタの分類精度を上げる方法を説明する。図12は、位置情報ログの集合例を示す説明図である。図12では、点線の矢印が位置情報ログに含まれる進行方向を示している。また、図12に示す例では、破線で囲まれた位置情報ログ群401に含まれる位置情報ログ401aと、位置情報ログ401bは、進行方向が同じであるが別の経路を利用していることを示す。また、図12に示す例では、破線で囲まれた位置情報ログ群402に含まれる位置情報ログ402aと、位置情報ログ402bは、同一移動経路上で大きく進行方向が変化していることを示す。本実施形態では、位置情報ログ群401に含まれる別経路を別の移動経路クラスタとして分類し、位置情報ログ群401に含まれる同一移動経路を同一の移動経路クラスタとして分類する精度を向上させる処理について説明する。
11 行動パタン解析基準入力部
20,50,60 行動パタン解析部
21 クラスタリング空間プロット部
22 滞留点クラスタ抽出部
23 滞留点位置情報ログ除去部
24 移動経路クラスタ抽出部
25 進行方向距離係数演算部
30 行動パタン記憶部
31 滞留点パタン保存部
32 移動経路パタン保存部
40 移動ベクトル演算解析部
41 移動ベクトル関数値演算部
42 滞留・移動経路属性判定部
Claims (10)
- 利用者の測位位置及び測位日時を含む情報である位置情報ログを、当該測位位置を表す数値情報および時刻により規定される多次元空間上にプロットする位置情報プロット手段と、
前記多次元空間において前記測位位置を表す数値情報により規定される空間である位置情報空間に対する時刻方向のユークリッド距離が近いと判定されやすくなる重み付けを前記位置情報ログに対して行い、重み付けされた位置情報ログをクラスタリングすることにより、利用者が頻繁に滞留する位置である滞留点を抽出する滞留点クラスタ抽出手段と、
位置情報プロット手段がプロットした位置情報ログから前記滞留点として抽出された位置情報ログを除いた位置情報ログの集合を非滞留点として抽出する非滞留点位置情報ログ抽出手段と、
前記位置情報空間に対する時刻方向のユークリッド距離が遠いと判定されやすくなる重み付けを前記非滞留点の位置情報ログに対して行い、重み付けされた位置情報ログをクラスタリングすることにより、利用者の移動経路を抽出する移動経路クラスタ抽出手段とを備えた
ことを特徴とする行動パタン解析装置。 - 位置情報プロット手段は、利用者の測位位置として緯度および経度を示す値を含む位置情報ログを、緯度、経度および時刻により規定される3次元空間上にプロットし、
滞留点クラスタ抽出手段は、前記3次元空間において緯度および経度により規定される平面である緯度経度平面に対する時刻方向のユークリッド距離が近いと判定されやすくなる重み付けを前記位置情報ログに対して行う
請求項1記載の行動パタン解析装置。 - 位置情報プロット手段は、利用者の測位位置として緯度、経度および高度を示す値を含む位置情報ログを、緯度、経度、高度および時刻により規定される4次元空間上にプロットし、
滞留点クラスタ抽出手段は、前記4次元空間において緯度、経度および高度により規定される空間である緯度経度高度空間に対する時刻方向のユークリッド距離が近いと判定されやすくなる重み付けを前記位置情報ログに対して行う
請求項1記載の行動パタン解析装置。 - 利用者の測位位置及び測位日時を含む情報である位置情報ログに含まれる当該測位日時順にソートされた隣接する2つの位置情報ログ間の差分値として、当該位置情報ログに含まれる前記測位日時の時間差、および、前記測位位置間の距離を算出する移動ベクトル関数値算出手段と、
前記測位日時の時間差及び測位位置間の距離に基づいて、前記位置情報ログの属性が、利用者が頻繁に滞留する位置である滞留点を示す滞留属性か、利用者の移動経路上の位置である非滞留点を示す非滞留属性かを判定する属性判定手段と、
前記位置情報ログを、前記測位位置を表す数値情報および時刻により規定される多次元空間上にプロットする位置情報プロット手段と、
前記多次元空間において前記測位位置を表す数値情報により規定される空間である位置情報空間に対する時刻方向のユークリッド距離が近いと判定されやすくなる重み付けを前記位置情報ログに対して行い、重み付けされた位置情報ログのうち前記滞留属性を有すると判定された位置情報ログをクラスタリングすることにより、前記滞留点を抽出する滞留点クラスタ抽出手段と、
前記位置情報空間に対する時刻方向のユークリッド距離が遠いと判定されやすくなる重み付けを前記位置情報ログに対して行い、重み付けされた位置情報ログのうち前記非滞留属性を有すると判定された位置情報ログをクラスタリングすることにより、利用者の移動経路を抽出する移動経路クラスタ抽出手段とを備えた
ことを特徴とする行動パタン解析装置。 - 移動ベクトル関数値算出手段は、2つの位置情報ログ間の差分値として、測位位置間での進行方向を表す進行方向角度の差分値を算出し、
位置情報プロット手段は、利用者の測位位置及び測位日時を含む位置情報ログに進行方向角度を加えたレコードを、測位位置を表す数値情報、時刻および前記進行方向角度により規定される多次元空間上にプロットし、
移動経路クラスタ抽出手段は、前記移動経路属性を有すると判定された位置情報ログに進行方向角度を加えたレコードをクラスタリングすることにより、利用者の移動経路を抽出する
請求項4記載の行動パタン解析装置。 - 2つの位置情報ログに基づいて、当該各位置情報ログが示す測位位置からの進行方向と、2つ位置情報ログが示す測位位置間を結ぶ方向とがなす角度の値が小さいほど値が小さくなる係数である進行方向距離係数を、前記位置情報ログごとに算出する進行方向距離係数算出手段を備え、
移動経路クラスタ抽出手段は、位置情報空間に対する時刻方向のユークリッド距離が遠いと判定されやすくなる重み付けを前記位置情報ログに対して行い、かつ、前記移動方向距離係数算出手段が算出した進行方向距離係数を前記ユークリッド距離に乗じることによる重み付けを行い、当該重み付けされた位置情報ログのうち非滞留属性を有すると判定された位置情報ログをクラスタリングすることにより、利用者の移動経路を抽出する
請求項4記載の行動パタン解析装置。 - 利用者の測位位置及び測位日時を含む情報である位置情報ログを、当該測位位置を表す数値情報および時刻により規定される多次元空間上にプロットし、
前記多次元空間において前記測位位置を表す数値情報により規定される空間である位置情報空間に対する時刻方向のユークリッド距離が近いと判定されやすくなる重み付けを前記位置情報ログに対して行い、重み付けされた位置情報ログをクラスタリングすることにより、利用者が頻繁に滞留する位置である滞留点を抽出し、
前記多次元空間上にプロットされた位置情報ログから前記滞留点として抽出された位置情報ログを除いた位置情報ログの集合を非滞留点として抽出し、
前記位置情報空間に対する時刻方向のユークリッド距離が遠いと判定されやすくなる重み付けを前記非滞留点の位置情報ログに対して行い、重み付けされた位置情報ログをクラスタリングすることにより、利用者の移動経路を抽出する
ことを特徴とする行動パタン解析方法。 - 利用者の測位位置及び測位日時を含む情報である位置情報ログに含まれる当該測位日時順にソートされた隣接する2つの位置情報ログ間の差分値として、当該位置情報ログに含まれる前記測位日時の時間差、および、前記測位位置間の距離を算出し、
前記測位日時の時間差及び測位位置間の距離に基づいて、前記位置情報ログの属性が、利用者が頻繁に滞留する位置である滞留点を示す滞留属性か、利用者の移動経路上の位置である非滞留点を示す非滞留属性かを判定し、
前記位置情報ログを、前記測位位置を表す数値情報および時刻により規定される多次元空間上にプロットし、
前記多次元空間において前記測位位置を表す数値情報により規定される空間である位置情報空間に対する時刻方向のユークリッド距離が近いと判定されやすくなる重み付けを前記位置情報ログに対して行い、重み付けされた位置情報ログのうち前記滞留属性を有すると判定された位置情報ログをクラスタリングすることにより、前記滞留点を抽出し、
前記位置情報空間に対する時刻方向のユークリッド距離が遠いと判定されやすくなる重み付けを前記位置情報ログに対して行い、重み付けされた位置情報ログのうち前記非滞留属性を有すると判定された位置情報ログをクラスタリングすることにより、利用者の移動経路を抽出する
ことを特徴とする行動パタン解析方法。 - コンピュータに、
利用者の測位位置及び測位日時を含む情報である位置情報ログを、当該測位位置を表す数値情報および時刻により規定される多次元空間上にプロットする位置情報プロット処理、
前記多次元空間において前記測位位置を表す数値情報により規定される空間である位置情報空間に対する時刻方向のユークリッド距離が近いと判定されやすくなる重み付けを前記位置情報ログに対して行い、重み付けされた位置情報ログをクラスタリングすることにより、利用者が頻繁に滞留する位置である滞留点を抽出する滞留点クラスタ抽出処理、
位置情報プロット処理でプロットされた位置情報ログから前記滞留点として抽出された位置情報ログを除いた位置情報ログの集合を非滞留点として抽出する非滞留点位置情報ログ抽出処理、および、
前記位置情報空間に対する時刻方向のユークリッド距離が遠いと判定されやすくなる重み付けを前記非滞留点の位置情報ログに対して行い、重み付けされた位置情報ログをクラスタリングすることにより、利用者の移動経路を抽出する移動経路クラスタ抽出処理
を実行させるための行動パタン解析プログラム。 - コンピュータに、
利用者の測位位置及び測位日時を含む情報である位置情報ログに含まれる当該測位日時順にソートされた隣接する2つの位置情報ログ間の差分値として、当該位置情報ログに含まれる前記測位日時の時間差、および、前記測位位置間の距離を算出する移動ベクトル関数値算出処理、
前記測位日時の時間差及び測位位置間の距離に基づいて、前記位置情報ログの属性が、利用者が頻繁に滞留する位置である滞留点を示す滞留属性か、利用者の移動経路上の位置である非滞留点を示す非滞留属性かを判定する属性判定処理、
前記位置情報ログを、前記測位位置を表す数値情報および時刻により規定される多次元空間上にプロットする位置情報プロット処理、
前記多次元空間において前記測位位置を表す数値情報により規定される空間である位置情報空間に対する時刻方向のユークリッド距離が近いと判定されやすくなる重み付けを前記位置情報ログに対して行い、重み付けされた位置情報ログのうち前記滞留属性を有すると判定された位置情報ログをクラスタリングすることにより、前記滞留点を抽出する滞留点クラスタ抽出処理、および、
前記位置情報空間に対する時刻方向のユークリッド距離が遠いと判定されやすくなる重み付けを前記位置情報ログに対して行い、重み付けされた位置情報ログのうち前記非滞留属性を有すると判定された位置情報ログをクラスタリングすることにより、利用者の移動経路を抽出する移動経路クラスタ抽出処理
を実行させるための行動パタン解析プログラム。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US13/977,386 US9497584B2 (en) | 2011-01-14 | 2012-01-12 | Action pattern analysis device, action pattern analysis method, and action pattern analysis program |
JP2012552685A JP5966930B2 (ja) | 2011-01-14 | 2012-01-12 | 行動パタン解析装置、行動パタン解析方法および行動パタン解析プログラム |
EP12734268.1A EP2664999A4 (en) | 2011-01-14 | 2012-01-12 | ACTIVITY DIAGRAM ANALYSIS UNIT, ACTIVITY DIAGRAM ANALYSIS METHOD, AND ACTIVITY DIAGRAM ANALYSIS PROGRAM |
Applications Claiming Priority (6)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2011006260 | 2011-01-14 | ||
JP2011-006260 | 2011-01-14 | ||
JP2011-109400 | 2011-05-16 | ||
JP2011109400 | 2011-05-16 | ||
JP2011-206345 | 2011-09-21 | ||
JP2011206345 | 2011-09-21 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2012096175A1 true WO2012096175A1 (ja) | 2012-07-19 |
Family
ID=46507082
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2012/000140 WO2012096175A1 (ja) | 2011-01-14 | 2012-01-12 | 行動パタン解析装置、行動パタン解析方法および行動パタン解析プログラム |
Country Status (4)
Country | Link |
---|---|
US (1) | US9497584B2 (ja) |
EP (1) | EP2664999A4 (ja) |
JP (1) | JP5966930B2 (ja) |
WO (1) | WO2012096175A1 (ja) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2016515230A (ja) * | 2013-02-06 | 2016-05-26 | フェイスブック,インク. | パターンのラベリング |
JP2016515229A (ja) * | 2013-02-06 | 2016-05-26 | フェイスブック,インク. | ルーチン逸脱の通知 |
JP2016537718A (ja) * | 2013-11-07 | 2016-12-01 | 華為技術有限公司Huawei Technologies Co.,Ltd. | 関心ポイントのクラスタリング方法及び関連する装置 |
JP2020107171A (ja) * | 2018-12-28 | 2020-07-09 | 株式会社日立製作所 | 行動パタン分析支援装置、行動パタン分析支援システム及び行動パタン分析支援方法 |
WO2020166064A1 (ja) * | 2019-02-15 | 2020-08-20 | 三菱電機株式会社 | 測位値分類装置 |
Families Citing this family (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8566961B2 (en) * | 2008-08-08 | 2013-10-22 | Absolute Software Corporation | Approaches for a location aware client |
JP6140490B2 (ja) * | 2013-03-19 | 2017-05-31 | 国立大学法人 東京大学 | 情報処理装置、情報処理方法及びプログラム |
US10114148B2 (en) * | 2013-10-02 | 2018-10-30 | Nec Corporation | Heterogeneous log analysis |
CN105589875B (zh) * | 2014-10-22 | 2019-10-25 | 方正国际软件(北京)有限公司 | 一种多轨迹绘制的方法及装置 |
US9282531B1 (en) * | 2015-03-02 | 2016-03-08 | Mitsubishi Electric Research Laboratories, Inc. | Method and system for localization of a device in an enclosed environment based on received signal strength levels |
KR101647078B1 (ko) * | 2015-06-23 | 2016-08-09 | 홍익대학교 산학협력단 | 자주 방문하는 장소를 판별하는 방법 및 이를 위한 장치 |
US10585933B2 (en) | 2017-08-16 | 2020-03-10 | International Business Machines Corporation | System and method for classification of low relevance records in a database using instance-based classifiers and machine learning |
CN111046895B (zh) * | 2018-10-15 | 2023-11-07 | 北京京东振世信息技术有限公司 | 一种确定目标区域的方法和装置 |
CN110290467B (zh) * | 2019-06-21 | 2020-08-04 | 清华大学 | 停留点、商业区服务范围、影响因素的获取方法及装置 |
CN110705510B (zh) * | 2019-10-16 | 2023-09-05 | 杭州优频科技有限公司 | 一种动作确定方法、装置、服务器和存储介质 |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2003076818A (ja) | 2001-08-31 | 2003-03-14 | Sony Corp | 行動管理装置、行動管理システム、及び行動管理方法 |
JP2009116541A (ja) * | 2007-11-05 | 2009-05-28 | Ntt Docomo Inc | 位置情報解析装置、情報配信システムおよび位置情報解析方法 |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20040016995A (ko) * | 2001-07-16 | 2004-02-25 | 웨이브마켓 인코포레이티드 | 무선 통신 네트워크에서 이동국들에 경보 기반 서비스들을제공하는 시스템 |
US6975939B2 (en) * | 2002-07-29 | 2005-12-13 | The United States Of America As Represented By The Secretary Of The Army | Mapping patterns of movement based on the aggregation of spatial information contained in wireless transmissions |
US8478747B2 (en) * | 2008-06-05 | 2013-07-02 | Samsung Electronics Co., Ltd. | Situation-dependent recommendation based on clustering |
US20100153292A1 (en) * | 2008-12-11 | 2010-06-17 | Microsoft Corporation | Making Friend and Location Recommendations Based on Location Similarities |
US8612134B2 (en) * | 2010-02-23 | 2013-12-17 | Microsoft Corporation | Mining correlation between locations using location history |
-
2012
- 2012-01-12 JP JP2012552685A patent/JP5966930B2/ja active Active
- 2012-01-12 WO PCT/JP2012/000140 patent/WO2012096175A1/ja active Application Filing
- 2012-01-12 US US13/977,386 patent/US9497584B2/en active Active
- 2012-01-12 EP EP12734268.1A patent/EP2664999A4/en not_active Withdrawn
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2003076818A (ja) | 2001-08-31 | 2003-03-14 | Sony Corp | 行動管理装置、行動管理システム、及び行動管理方法 |
JP2009116541A (ja) * | 2007-11-05 | 2009-05-28 | Ntt Docomo Inc | 位置情報解析装置、情報配信システムおよび位置情報解析方法 |
Non-Patent Citations (4)
Title |
---|
HIDEKI YOSHII ET AL.: "A Study on analyzing GPS-based Location Data Using Segment Velocity Technique", IEICE TECHNICAL REPORT, vol. 109, no. 39, 14 May 2009 (2009-05-14), pages 109 - 112, XP008170509 * |
KAZUYA OKADA ET AL.: "Proposal of a Method for Extracting Points of Interest from User's Trajectory", IPSJ SIG NOTES, vol. 2008, no. 91, 26 September 2008 (2008-09-26), pages 89 - 94, XP008171039 * |
See also references of EP2664999A4 |
SHINJI KAMI ET AL.: "Algorithm for Detecting Significant Locations from Raw GPS Data", 2010 NEN IEICE COMMUNICATIONS SOCIETY CONFERENCE KOEN RONBUNSHU, vol. 6332, 31 August 2010 (2010-08-31), pages 221 - 235, XP019155481 * |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2016515230A (ja) * | 2013-02-06 | 2016-05-26 | フェイスブック,インク. | パターンのラベリング |
JP2016515229A (ja) * | 2013-02-06 | 2016-05-26 | フェイスブック,インク. | ルーチン逸脱の通知 |
US10585961B2 (en) | 2013-02-06 | 2020-03-10 | Facebook, Inc. | Pattern labeling |
JP2016537718A (ja) * | 2013-11-07 | 2016-12-01 | 華為技術有限公司Huawei Technologies Co.,Ltd. | 関心ポイントのクラスタリング方法及び関連する装置 |
US10423728B2 (en) | 2013-11-07 | 2019-09-24 | Huawei Technologies Co., Ltd. | Clustering method for a point of interest and related apparatus |
JP2020107171A (ja) * | 2018-12-28 | 2020-07-09 | 株式会社日立製作所 | 行動パタン分析支援装置、行動パタン分析支援システム及び行動パタン分析支援方法 |
JP7343973B2 (ja) | 2018-12-28 | 2023-09-13 | 株式会社日立製作所 | 行動パタン分析支援装置、行動パタン分析支援システム及び行動パタン分析支援方法 |
WO2020166064A1 (ja) * | 2019-02-15 | 2020-08-20 | 三菱電機株式会社 | 測位値分類装置 |
JPWO2020166064A1 (ja) * | 2019-02-15 | 2021-05-20 | 三菱電機株式会社 | 測位値分類装置 |
JP7076589B2 (ja) | 2019-02-15 | 2022-05-27 | 三菱電機株式会社 | 測位値分類装置 |
Also Published As
Publication number | Publication date |
---|---|
JPWO2012096175A1 (ja) | 2014-06-09 |
JP5966930B2 (ja) | 2016-08-10 |
EP2664999A1 (en) | 2013-11-20 |
US20130273937A1 (en) | 2013-10-17 |
US9497584B2 (en) | 2016-11-15 |
EP2664999A4 (en) | 2015-01-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2012096175A1 (ja) | 行動パタン解析装置、行動パタン解析方法および行動パタン解析プログラム | |
CN109003028B (zh) | 用于划分物流区域的方法和装置 | |
US9392463B2 (en) | System and method for detecting anomaly in a handheld device | |
Soh et al. | Adaptive deep learning-based air quality prediction model using the most relevant spatial-temporal relations | |
US10133938B2 (en) | Apparatus and method for object recognition and for training object recognition model | |
US10748121B2 (en) | Enriching calendar events with additional relevant information | |
US9208220B2 (en) | Method and apparatus of text classification | |
US20140164390A1 (en) | Mining trajectory for spatial temporal analytics | |
US20120148102A1 (en) | Mobile body track identification system | |
Koo et al. | PDR/fingerprinting fusion indoor location tracking using RSS recovery and clustering | |
CN110533055A (zh) | 一种点云数据的处理方法和装置 | |
JP2014132711A (ja) | 複数の測位方式に基づく移動端末の位置情報から対象エリアを特定するエリア管理サーバ、プログラム及び方法 | |
CN111475746B (zh) | 兴趣点位置挖掘方法、装置、计算机设备和存储介质 | |
CN110730207B (zh) | 基于位置服务的位置识别方法、装置、设备及存储介质 | |
US11006238B1 (en) | Method for profiling based on foothold and terminal using the same | |
CN111476059A (zh) | 目标检测方法、装置、计算机设备及存储介质 | |
CN107038593B (zh) | 一种基于防伪溯源***的异常数据处理方法及*** | |
CN116610849A (zh) | 获取轨迹相似的移动对象的方法、装置、设备及存储介质 | |
CN111667920B (zh) | 一种基于定位数据的人群密切接触排查方法 | |
Alim et al. | Sampling-based estimation method for parameter estimation in big data business era | |
CN111563137B (zh) | 一种用于重合轨迹的分析方法和*** | |
CN113284337A (zh) | 基于车辆轨迹多维数据的od矩阵计算方法及装置 | |
WO2012169152A1 (ja) | 移動目的地予測装置、移動目的地予測方法および移動目的地予測プログラム | |
Gupta et al. | Svad: End-to-end sensory data analysis for iobt-driven platforms | |
Arslan et al. | Using spatio-temporal trajectories to monitor construction sites for safety management |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 12734268 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2012552685 Country of ref document: JP Kind code of ref document: A |
|
REEP | Request for entry into the european phase |
Ref document number: 2012734268 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2012734268 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 13977386 Country of ref document: US |
|
NENP | Non-entry into the national phase |
Ref country code: DE |