CN103995859B - A kind of hot spot region incident detection system based on geographical labels applied to LBSN networks - Google Patents

Abstract

The invention discloses a kind of detection system of the hot spot region event based on geographical labels applied to LBSN networks, the detection system is operated in LBSN, belongs to network data processing technique；The detection system is made up of cluster module of registering, the area calculation module based on label clustering and hot spot region event computing module.Cluster module of registering is used to obtain the corresponding affiliated geographic area of information of registering to information progress clustering processing of registering；Area calculation module based on label clustering obtains cluster inner region set using geographical labels clustering algorithm from geographic area belonging to information of registering is corresponding；Frequency of registering in the event computing module application time window of hot spot region obtains hot spot region event from cluster inner region set, so that the hot spot region event of acquisition is supplied into user.The detection system for the hot spot region event based on geographical labels that the present invention is designed is clustered in smaller scope using cluster to the point in cluster further, advantage of this is that can greatly reduce the data volume calculated in LBSN, improves computational efficiency.

Description

A kind of hot spot region incident detection based on geographical labels applied to LBSN networks System

Technical field

The present invention relates to a kind of technical field of registering of geographical labels, more particularly, refer to a kind of applied to LBSN nets The hot spot region incident detection system based on geographical labels of network, wherein hot spot region are the clusters based on label and geographical position The division of progress.

Background technology

Geographical labels (Geo Tags) refer to the data message for describing geographical position residing for point of interest, its information content Include point of interest address information, point of interest latitude and longitude information.Geographical labels can preferably digitize the geographical position of point of interest Confidence ceases, and is conducive to global metadata positioning and geographical location information to review.Geographical labels are also referred to as geographical indication.Record of registering is Refer to the main body of the society and carry out the obtained data message of registering in point of interest.

At present, location-based social networks (LBSN Location-based Social Networking) is increasingly It is popular.Due to the fast development of quick forth generation mobile communications network, and to Map Services and the intelligent hand of embedded GPS module The powerful interface of machine supports that it is easily mobile subscriber and recognizes their position, and shares their LBSN databases.One In individual LBSN databases, user can be found that and created point of interest (poi point of interest), can working as at them Front position is registered, and is made comments and opinion and addition good friend etc..Therefore, LBSN networks such as Foursquare, Facebook Places, Sina weibo etc., have had taken up different mechanism to attract user, and encourage user to share them Information of registering.Also, having had some researchs to start with these has what user produced to be registered information with geographical labels.Cause For these data researcher can be allowed to go to analyze the benefits program of social hierarchy in the way of data-driven, and according to registering The Move Mode of INFORMATION DISCOVERY user, predicts friend relation, is better understood from the different aspect in city.Letter of registering can also be utilized Cease to find hot spot region.

It is currently based on the hot spot region event containing geographical labels and finds mainly there is a kind of mode：Artificial division is well first Grid is managed, the information of registering on respective region is then counted, total amount of registering reaches that certain threshold value is then designated hot spot region.It is this Method has three, and (1) first, advance zoning may carry out actual region to be divided into different grids, it is impossible to Reflect actual hot spot region.(2) criterion of hot spot region is whether the total quantity of registering of the advance zoning reaches One threshold value, reaches, is designated hot spot region, but do not account for the influence of time factor.(3) region of mesh generation is compared Greatly, it is difficult to the more accurate region of positioning.

The content of the invention

The feature for data of being registered for LBSN, and the deficiency of processing method, this hair are found to existing hot spot region event It is bright to propose a kind of hot spot region incident detection system based on geographical labels.The hot spot region incident detection system considers User builds coarseness region division with reference to geospatial information, then, using geography in recent historical record of registering The clustering algorithm of label calculates fine granularity regional extent, finally calculates the hot zone under certain time window in region Domain.The hot spot region incident detection system based on geographical labels that the present invention is designed is embedded in LBSN databases, followed by The LBSN network operations.

A kind of hot spot region incident detection system based on geographical labels applied to LBSN networks that the present invention is designed, institute State and set the hot spot region event based on geographical labels to visit between the LBSN databases (2) in LBSN networks and user (1) Examining system (3)；

The hot spot region incident detection system (3) based on geographical labels includes cluster module of registering (31), is based on The area calculation module (32) and hot spot region event computing module (33) of label clustering；The hot spot region event computing module (33) it is to be connected interface between LBSN databases (2) and user (1)；

Cluster module (31) first aspect of registering is used to send the request of registering containing geographical labels to LBSN databases (2) Information Q_31-2, the Q_31-2=R_POI_p(x,y),POI；

R_POI_p(x, y) represents sign-in desk geographical position, and x represents longitude, and y represents latitude；

POI represents geographical labels；Any one geographical labels in the POI are designated as a, and another geographical labels is designated as b, a,b∈POI；

Cluster module (31) second aspect of registering is according to Q_31-2=R_POI_p(x, y), POI can be in LBSN databases (2) The record of registering matched with geographical labels POI is searched out, the return information Q that registers is designated as_2-31；

Cluster module (31) third aspect of registering is to the return information Q that registers that receives_2-31According to cluster interval time Kcluster-span carries out k-means clustering method processing, obtains region block message Q_31-32, the Q_31-32={ regX₁, regX₂,…,regX_y, then by Q_31-32Export to the area calculation module (32) based on label clustering；

regX₁Represent first region unit in the R of any one geographic area；

regX₂Represent second region unit in the R of any one geographic area；

regX_yRepresent last region unit in the R of any one geographic area；

Y represents region unit number；

Area calculation module (32) first aspect based on label clustering is used for receiving area block message Q_31-32={ regX₁, regX₂,…,regX_y}；

Area calculation module (32) second aspect based on label clustering is tactful POI-CP pairs according to geographical labels cluster Q_31-32={ regX₁,regX₂,…,regX_yHandled, obtain convergence geographical labels region unit Q_32-2；And will the geographical mark of convergence Label region unit is written to LBSN databases (2)；

Hot spot region event computing module (33) first aspect receives the hot spot region inquiry request Request of user (1), The Request={ Geo (x, y), dist, Hot }, and Request={ Geo (x, y), dist, Hot } is transmitted to LBSN numbers According to storehouse (2)；

Hot spot region event computing module (33) second aspect can according to Request={ Geo (x, y), dist, Hot } The hot spot region matched with Geo (x, y) is searched out in LBSN databases (2), inquiry return information Q is designated as_2-33；

Hot spot region event computing module (33) third aspect is according to POI-TP pairs of the frequency strategy of registering under time window The Q_2-33Calculating processing is carried out, region focus incident and ranking ChecFreq is obtained, and the ChecFreq is fed back into use Family (1).

In the present invention, described geographical labels cluster strategy POI-CP has the following steps：

Extraction belongs to the same area block regX_yIn geographical labels POI the step of；

Calculating belongs to the same area block regX_yIn geographical labels POI position numberThe step of；

Calculate geographical labels POI and the place-centric point of geographical labels position Between maximum linear distanceThen judge describedWith zone radius threshold value r_{Threshold value}Size, ifThen by r_{Threshold value}It is assigned to the region unit distance correlation radius in affiliated areaIfThen choose maximum linear distance and be used as the region unit distance correlation radius in affiliated areaAnd then the distance correlation radius for passing through a geographical labelsWith the distance correlation radius of b geographical labelsSum is than the central point distance in upper geographical labels POI between any two geographical labels a, bObtain Distance correlationThe step of；

Calculate the semantic dependency between any two geographical labels a, b in geographical labels POIThe step of；

According to describedWith it is describedWith distance correlation threshold value rel_Distance, it is semantic related Property threshold value rel_{It is semantic}Contrasted, and according to comparing result combined region block regX_yThe step of；

IfAndWhen, by the position of registering of b geographical labelsIt is merged into the position of registering of a geographical labels

IfOrWhen, then the position of registering of b geographical labelsNot with the position of registering of a geographical labelsCarry out region merged block.

In the present invention, the concrete mode of the frequency strategy POI-TP that registers under described time window is：Hot spot region Event computing module (33) can calculate user (1) in real time and ask any one region regX_yFocus incident, when poly- to hot spot region Class database (2) asks the history in respective region by the end of current time t to contain geography and register and record Q_2-33, then event description ForΔ T represents time window, Δ T=| t- (t-1) |, when t represents current Carve, t-1 represents previous moment, ChecFreq^tRepresent the frequency of registering at current time t, ChecFreq^t-1Represent previous moment t-1 Frequency of registering；

The active degree Rank and Δ T of event register frequency and lasting event of registering is directly proportional, i.e.,：

WhereinRepresent the frequency of registering in time window Δ T, regX_yAny one region is represented, that is, is summed Element, i represents summing target, Ω_iRepresent that user (1) asks scope inner region to be registered total quantity in time window Δ T,Represent the time window number in current all hot spot regionsMaximum, j represents time window The maximum area identification number of number.

The advantage for the hot spot region incident detection system based on geographical labels that the present invention is designed is：

1. solved and made using artificial division region using incident detection system in hot spot region of the present invention in LBSN databases Into the low defect of hot spot region precision.Incident detection system in hot spot region of the present invention is first in the ground on Label space to focus Region is clustered, and is obtained rough hot spot region and is divided, reduces the data volume in cluster process.

2. incident detection system in hot spot region of the present invention is obtained and true using fine-grained excavation is carried out in coarseness cluster The hot spot region being consistent so that the hot spot region event of acquisition finds more preferably to meet actual demand.

3. incident detection system in hot spot region of the present invention divides hot spot region using time window, passes through rate of change threshold of registering Value searches out hot spot region so that the query time of LBSN networks shortens, and improves response speed.

4. incident detection system in hot spot region of the present invention uses modular design method, and LBSN networks pass through hot zone Domain event computing module is linking interface, is realized and user mutual, improves the efficiency that user inquires about hot spot region event.

Brief description of the drawings

Fig. 1 is the structured flowchart of the hot spot region incident detection system of the invention based on geographical labels.

Fig. 2 is the timing diagram of the hot spot region incident detection system of the invention based on geographical labels.

Embodiment

Below in conjunction with accompanying drawing, the present invention is described in further detail.

The hot spot region incident detection system architecture diagram registered based on user according to the present invention is given in Fig. 1. The hot spot region incident detection system based on geographical labels for setting the present invention to design between existing LBSN databases 2 and user 1 3, the system includes cluster module 31 of registering, the area calculation module 32 based on label clustering and hot spot region event and calculates mould Block 33.The hot spot region event computing module 33 is to be connected interface between LBSN databases 2 and user 1.

In the present invention, location-based social networks (LBSN Location-based Social are utilized Networking) constitute LBSN databases 2 provide history register information carry out hot spot region incident detection, be to search out The data source of hot spot region related information.

In the present invention, the information of registering of the point of interest POI in any one geographic area is designated as R_POI_p(x,y),POI。R Geographic area is represented, POI represents the geographical labels of the character string in R, the i.e. point of interest of place geographic area, be also semantic point Source file string needed for analysis, POI character length is designated as L_POI, p represents the number of times that the main body of the society is registered in POI, POI_p(x, Y) the sign-in desk geographical position of pth time is represented, x represents longitude, and y represents latitude.

Usually, for the statement of present patent application content, geographical labels POI could be arranged in any one region R Dining room A, hospital B, library C, teaching building D etc.；Geographical labels POI is expressed as POI={ A, B, C, D } using aggregate form, is Generalized list reaches any one geographical labels in geographical labels POI, the POI and is designated as a, and another geographical labels is designated as b, a, b ∈POI.Being described as based on geographical labels：

Dining room A in the R of any one geographic area information of registering is designated as R_A_α(x,y),A；R represents geographic area, and A is represented Character string in R, A character length is designated as L_A(i.e. " geographic area " " dining room ", L_A=12,2 bytes of a Chinese character), a tables Show the number of times that the main body of the society is registered in A, A_α(x, y) represents sign-in desk geographical position, and x represents longitude, and y represents latitude.

Hospital B in the R of any one geographic area information of registering is designated as R_B_β(x,y),B；R represents geographic area, and B is represented Character string in R, B character length is designated as L_B(i.e. " geographic area " " hospital ", L_A=12), β represents that the main body of the society is signed in B The number of times arrived, B_β(x, y) represents sign-in desk geographical position, and x represents longitude, and y represents latitude.

Library C in the R of any one geographic area information of registering is designated as R_C_γ(x,y),C；R represents geographic area, C tables Show the character string in R, C character length is designated as L_C(i.e. " geographic area " " library ", L_A=14), γ represents the main body of the society The number of times registered in C, C_γ(x, y) represents sign-in desk geographical position, and x represents longitude, and y represents latitude.

Teaching building D in the R of any one geographic area information of registering is designated as R_D_θ(x,y),D；R represents geographic area, D tables Show the character string in R, D character length is designated as L_D(i.e. " geographic area " " teaching building ", L_A=14), θ represents that the main body of the society exists The number of times that D registers, D_θ(x, y) represents sign-in desk geographical position, and x represents longitude, and y represents latitude.

User 1

User 1 is in the hot spot region incident detection system 3 based on geographical labels designed using the present invention, first aspect Interest request Request={ Geo (x, y), dist, Hot } is sent to hot spot region event computing module 33；Second aspect is used In the cluster hot spot area in real time information ChecFreq for receiving the return of hot spot region event computing module 33.

Geo (x, y) in the recommendation request Request={ Geo (x, y), dist, Hot } is represented where request user Geographical position, x is longitude, and y is latitude；Dist represents the interest distance radius that user is set；Hot represents the heat of user's concern Point region.

Referring to shown in Fig. 1 and Fig. 2, in the present invention, the hot spot region incident detection system 3 based on geographical labels includes Register cluster module 31, the area calculation module 32 based on label clustering and hot spot region event computing module 33.Below will be detailed Describe bright modules in detail：

Register cluster module 31

The first aspect of cluster module 31 of registering, which is used to sending the user containing geographical labels to LBSN databases 2, registers information Q_31-2=R_POI_p(x,y),POI；

The second aspect of cluster module 31 of registering is according to Q_31-2=R_POI_p(x, y), POI can be searched in LBSN databases 2 Go out multiple record Q that register matched with geographical labels POI_2-31；

The third aspect of cluster module 31 of registering is registered information Q to the return received_2-31According to cluster interval time Kcluster-span carries out k-means clustering method processing, obtains region block message Q_31-32={ regX₁,regX2_,…, regX_y, then by Q_31-32Export to the area calculation module 32 based on label clustering.

In the present invention, the Q_31-2=R_POI_pR_POI in (x, y), POI_p(x, y) represents sign-in desk geographical position, X represents longitude, and y represents latitude.POI represents geographical labels, i.e. R_POI_pGeographic name where (x, y), is also source file string The content of record；The POI={ A, B, C, D }, A is dining room geographical labels, and B is hospital's geographical labels, and C is library's geography mark Label, D is teaching building geographical labels.

In the present invention, region block message Q_31-32={ regX₁,regX₂,…,regX_yIn regX₁Represent any one First region unit in the R of geographic area, regX₂Represent second region unit in the R of any one geographic area, regX_yRepresent Last region unit in the R of any one geographic area, y represents region unit number.regX_yIt is also referred to as geographical at any one Any one region unit marked off in the R of region.

Enumerate, meeting dining room A record of registering has R_A₁(x,y),A、R_A₂(x,y),A、R_A₃(x,y),A、R_A₄(x, y),A、R_A₅(x,y),A、……、R_A_α(x,y),A；

R_A₁(x, y) represents dining room A first sign-in desk position；

R_A₂(x, y) represents dining room A second sign-in desk position；

R_A₃(x, y) represents dining room A the 3rd sign-in desk position；

R_A₄(x, y) represents dining room A the 4th sign-in desk position；

R_A₅(x, y) represents dining room A the 5th sign-in desk position；

R_A_α(x, y) represents dining room A last sign-in desk position；For convenience of explanation, R_A_α(x, y) is also referred to as food Hall A any one sign-in desk position；

Enumerate, meeting hospital B record of registering has R_B₁(x,y),B、R_B₂(x,y),B、R_B₃(x,y),B、R_B₄(x, y),B、R_B₅(x,y),B、……、R_B_β(x,y),B；

R_B₁(x, y) represents hospital B first sign-in desk position；

R_B₂(x, y) represents hospital B second sign-in desk position；

R_B₃(x, y) represents hospital B the 3rd sign-in desk position；

R_B₄(x, y) represents hospital B the 4th sign-in desk position；

R_B₅(x, y) represents hospital B the 5th sign-in desk position；

R_B_β(x, y) represents hospital B last sign-in desk position；For convenience of explanation, R_B_β(x, y) is also referred to as doctor Institute B any one sign-in desk position；

Enumerate, meeting library C record of registering has R_C₁(x,y),C、R_C₂(x,y),C、R_C₃(x,y),C、R_C₄ (x,y),C、R_C₅(x,y),C、……、R_C_γ(x,y),C；

R_C₁(x, y) represents library C first sign-in desk position；

R_C₂(x, y) represents library C second sign-in desk position；

R_C₃(x, y) represents library C the 3rd sign-in desk position；

R_C₄(x, y) represents library C the 4th sign-in desk position；

R_C₅(x, y) represents library C the 5th sign-in desk position；

R_C_γ(x, y) represents library C last sign-in desk position；For convenience of explanation, R_C_γ(x, y) is also referred to as Library C any one sign-in desk position；

Enumerate, meeting teaching building D record of registering has R_D₁(x,y),D、R_D₂(x,y),D、R_D₃(x,y),D、R_D₄ (x,y),D、R_D₅(x,y),D、……、R_D_θ(x,y),D；

R_D₁(x, y) represents teaching building D first sign-in desk position；

R_D₂(x, y) represents teaching building D second sign-in desk position；

R_D₃(x, y) represents teaching building D the 3rd sign-in desk position；

R_D₄(x, y) represents teaching building D the 4th sign-in desk position；

R_D₅(x, y) represents teaching building D the 5th sign-in desk position；

R_D_θ(x, y) represents teaching building D last sign-in desk position；For convenience of explanation, R_D_θ(x, y) is also referred to as Teaching building D any one sign-in desk position.

According to dining room A, hospital B, library C, the teaching building D illustrated out, then the record of registering matched with geographical labels POI Q_2-31；

Enumerate, first region unit regX₁Including record of registering have R_A₂(x,y),A、R_A₃(x,y),A、R_A₄(x, y),A、R_A₅(x,y),A、R_A_α(x,y),A、R_B₁(x,y),B、R_B₂(x,y),B、R_B₃(x,y),B、R_B₄(x,y),B、R_ C₁(x,y),C、R_C₂(x,y),C、R_D₁(x, y), D and R_D_θ(x,y),D；First region unit regX₁Using aggregate form table Up to for：

Enumerate, second region unit regX₂Including record of registering have R_A₁(x,y),A、R_B₅(x,y),B、R_B_β(x, y),B、R_C₃(x,y),C、R_C₄(x,y),C、R_C₅(x, y), C and R_C_γ(x,y),C；Second region unit regX₂Using collection Conjunction form is expressed as：

Enumerate, last region unit regX_yIncluding record of registering have R_D₂(x,y),D、R_D₃(x,y),D、R_D₄ (x, y), D and R_D₅(x,y),D；Last region unit regX_yIt is expressed as using aggregate form：

Regional block is recorded, uses aggregate form to express region block message for Q_31-32={ regX₁,regX₂,…, regX_y}。

In the present invention, k-means clustering methods refer to《At big data internet large-scale data excavates and is distributed Reason》, Wang Binyi, September the 1st edition in 2012.

In the present invention, cluster module 31 of registering is set to simple task, is in order among huge LBSN databases 2 Rough record of registering is obtained, rough record application k-means clustering methods of registering are handled, can be by record aggregate of registering To respective region.

Use the R_POI in k-means clustering method processing datas storehouse_p(x, y), does not provide externally and handles clothes on line in real time Business, the purpose for cluster module 31 of registering is the scope (zoning for being referred to as coarseness) for substantially marking off region CR；In addition, Region CR scope spatially changes not substantially registering, can according to cluster interval time kcluster-span it is lower online at Manage region CR division.

Area calculation module 32 based on label clustering

The first aspect of area calculation module 32 based on label clustering is used for receiving area block message Q_31-32={ regX₁, regX₂,…,regX_y}；

The second aspect of area calculation module 32 based on label clustering clusters strategy POI-CP to Q according to geographical labels_31-32 ={ regX₁,regX₂,…,regX_yHandled, obtain convergence geographical labels region unit Q_32-2；And will convergence geographical labels area Domain block Q_32-2It is written to LBSN databases 2.

In the present invention, specific processing of the explanation to geographical labels, i.e. geographical labels cluster strategy POI-CP is set forth below Implementation steps：

(1) first area block

In the present invention, to first region unit regX₁In geographical labels carry out geographical labels cluster strategy POI-CP Process step be：

Step 101：Extract the geographical labels belonged in the same area block

From Q_31-32={ regX₁,regX₂,…,regX_yIn extract satisfaction first region unit regX₁Geographical labels, If first region unit regX₁In geographical labels include A, B and C, then meet regX₁Geographical labels use set description For；

Step 102：The position number of geographical labels is obtained；

To first region unit regX₁In geographical labels A carry out position classification, be met regX₁In A geography mark Sign positionThe A is in regX₁Time of middle appearance Number scale isAnd

To first region unit regX₁In geographical labels B carry out position classification, be met regX₁In B geography mark Sign positionThe B is in regX₁Time of middle appearance Number scale isAnd

To first region unit regX₁In geographical labels C carry out position classification, be met regX₁In C geography mark Sign positionThe C is in regX₁The number of times of middle appearance is designated as And

To first region unit regX₁In geographical labels D carry out position classification, be met regX₁In D geography mark Sign positionThe D is in regX₁The number of times of middle appearance is designated as And

In the present invention, statistics is in first region unit regX₁In all geographical labels number of times of registering, be designated asAnd

Step 103：Distance correlation

Step 103-1：ParsingIn warp Spend average valueWith latitude average valueI represents asking in summation relation And index,It is met regX₁In A geographical labels position place-centric pointParsingIn each label point position arrive Distance, and select maximum linear distance, be designated as

ParsingIn longitude average valueWith latitude average valueJ represents that the summation in summation relation refers to Mark,It is met regX₁In B geographical labels position place-centric pointParsingIn each label point position arrive Distance, and select maximum linear distance and be designated as

ParsingIn longitude average valueWith Latitude average valueM represents the summing target in summation relation,Obtain AC geography marks Sign the place-centric point of positionParsing In each label point position arriveDistance, and select maximum linear distance and be designated as

ParsingIn longitude average valueAnd latitude Spend average valueN represents the summing target in summation relation,Obtain AD geographical labels The place-centric point of positionParsingIn Each label point position is arrivedDistance, and select maximum linear distance and be designated as

In the present invention, statistics is in first region unit regX₁In all geographical labels and geographical labels position position Put central pointBetween maximum linear distanceAnd

Step 103-2：Setting area block radius threshold is designated as r_{Threshold value}；

If maximum linear distance is less than zone radius threshold value r_{Threshold value}, then by r_{Threshold value}It is assigned to the region unit distance in affiliated area Correlation radius

If maximum linear distance is more than or equal to zone radius threshold value r_{Threshold value}, then choose maximum linear distance and be used as affiliated area In region unit distance correlation radius

It can similarly obtain：The distance correlation radius of a geographical labels is designated asThe distance correlation of b geographical labels half Footpath is designated as

Enumerate, ifThen by r_{Threshold value}It is assigned to and meets regX₁In region unit distance correlation radiusIfThen willIt is assigned to and meets regX₁In region unit distance correlation radius

Enumerate, ifThen by r_{Threshold value}It is assigned to and meets regX₁In region unit distance correlation radiusIfThen willIt is assigned to and meets regX₁In region unit distance correlation radius

Enumerate, ifThen by r_{Threshold value}It is assigned to and meets regX₁In region unit distance correlation radiusIfThen willIt is assigned to and meets regX₁In region unit distance correlation radius

Enumerate, ifThen by r_{Threshold value}It is assigned to and meets regX₁In region unit distance correlation radiusIfThen willIt is assigned to and meets regX₁In region unit distance correlation radius

Step 103-3：Calculating meets regX₁In any two geographical labels position central point distance；

WithCentral point distance be designated as

WithCentral point distance be designated as

WithCentral point distance be designated as

WithCentral point distance be designated as

WithCentral point distance be designated as

In the present invention, statistics is in first region unit regX₁In geographical labels POI in any two geographical labels a, Central point distance between b is designated as

Step 103-4：Definition meets regX₁In distance correlation

Enumerate,WithDistance correlation be designated as

Enumerate,WithDistance correlation be designated as

Enumerate,WithDistance correlation be designated as

Enumerate,WithDistance correlation be designated as

Enumerate,WithDistance correlation be designated as

In the present invention, plan range is that known range formula calculating is obtained, such as

In the present invention, statistics is in first region unit regX₁In distance correlation be designated as

Step 104：Semantic dependency

In the present invention, it is the editing distance between POI, i.e. E to define semantic distance_POI.The editing distance E_POIFor handle Source file string POI={ A, B, C, D } is converted into the minimum sequence of operation of target strings TPOI={ A, B, C, D } cost.Editing distance Calculating refer to the 14th printing December the 1st edition in 2009《Introduction to algorithms》The 218-219 pages, (U.S.) Thomas H.Cormen Charles E.Leiserson Ronald L.Rivest Clifford Stein write, Pan Jingui, Gu Tiecheng, Li Chengfa, Ye Maoyi.

Enumerate, in first region unit regX₁In dining room A string length be designated asHospital B character string is long Degree is designated asLibrary C string length is designated asTeaching building D string length is designated as

Enumerate, in first region unit regX₁In dining room A and hospital B editing distance be designated asDining room A and figure Book shop C editing distance is designated asDining room A and teaching building D editing distance are designated asHospital B and library C volume Collecting distance isHospital B and teaching building D editing distance are designated as

Enumerate, in first region unit regX₁In dining room A and hospital B semantic dependency be designated as

Enumerate, in first region unit regX₁In dining room A and library C semantic dependency be designated as

Enumerate, in first region unit regX₁In dining room A and teaching building D semantic dependency be designated as

Enumerate, in first region unit regX₁In hospital B and library C semantic dependency be designated as

Enumerate, in first region unit regX₁In hospital B and teaching building D semantic dependency be designated as

In the present invention, in area information Q_31-32In the semantic dependencies of any two geographical labels be designated as

Step 105：Whether region unit merges

Distance correlation threshold value is set to be designated as rel_Distance, semantic dependency threshold value be designated as rel_{It is semantic}, and according to rel_DistanceWith rel_{It is semantic}Whether the merging treatment of region unit is carried out；

Step 105-1：IfAndWhen, willIt is merged intoIn, thenIt is updated toAnd willExport and give LBSN data Storehouse, performs step 105-2；

Step 105-2：IfOrWhen, thenWith Without region merged block；And willWithExport and give LBSN databases, perform step 105-3；

Step 105-3：IfAndWhen, willIt is merged intoIn, thenIt is updated toAnd willExport and give LBSN numbers According to storehouse, step 105-4 is performed；

Step 105-4：IfOrWhen, thenWith Without region merged block；And willWithExport and give LBSN databases, perform step 105-5；

If step 105-5AndWhen, willIt is merged intoIn, thenIt is updated toAnd will

Step 105-6：IfOrWhen, thenWith Without region merged block；And willWithExport and give LBSN databases, perform step 105-7；

Step 105-7：IfAndWhen, willIt is merged intoIn, thenIt is updated toAnd willExport and give LBSN numbers According to storehouse, step 105-8 is performed；

Step 105-8：IfOrWhen, thenWith Without region merged block；And willWithExport and give LBSN databases, perform step 105-9；

Step 105-9：IfAndWhen, willIt is merged intoIn, thenIt is updated toAnd willExport and give LBSN data Storehouse, performs step 105-10；

Step 105-10：IfOrWhen, thenWithWithout region merged block；And willWithExport and give LBSN databases.

In the present invention, by first region unit regX₁Write-in LBSN databases 2 register dot position information using collection Conjunction form is expressed as

(2) second area block

In the present invention, to second region unit regX₂In geographical labels carry out geographical labels cluster strategy POI-CP Process step be：

Step 201：Extract the geographical labels belonged in the same area block

From Q_31-32={ regX₁,regX₂,…,regX_yIn extract satisfaction second region unit regX₂Geographical labels, If second region unit regX₂In geographical labels include B and C, then meet regX₂Geographical labels use set description for；

Step 202：The position number of geographical labels is obtained；

To second region unit regX₂In geographical labels B carry out position classification, be met regX₂In B geography mark Sign positionThe B is in regX₂The number of times of middle appearance is designated as And

To second region unit regX₂In geographical labels C carry out position classification, be met regX₂In C geography mark Sign positionThe C is in regX₂Time of middle appearance Number scale isAnd

In the present invention, statistics is in second region unit regX₂In all geographical labels number of times of registering, be designated asAnd

Step 203：Distance correlation

Step 203-1：ParsingIn longitude average valueWith latitude average valueJ represents that the summation in summation relation refers to Mark,It is met regX₂In B geographical labels position place-centric pointSolution AnalysisIn each label point position arriveAway from From, and maximum linear distance is selected, it is designated as

ParsingIn longitude average valueWith latitude average valueM represents the summing target in summation relation, m ∈ γ ', are met regX₂In C geographical labels position place-centric pointParsingIn each label point position arrive Distance, and select maximum linear distance, be designated as

In the present invention, statistics is in second region unit regX₂In all geographical labels and geographical labels position position Put central pointBetween maximum linear distanceAnd

Step 203-2：Setting area block radius threshold is designated as r_{Threshold value}；

If maximum linear distance is less than zone radius threshold value r_{Threshold value}, then by r_{Threshold value}It is assigned to the region unit distance in affiliated area Correlation radius rD_POI；

If maximum linear distance is more than or equal to zone radius threshold value r_{Threshold value}, then maximum linear distance is chosen in affiliated area Region unit distance correlation radius rD_POI；

It can similarly obtain：The distance correlation radius of a geographical labels is designated asThe distance correlation of b geographical labels half Footpath is designated as

Enumerate, ifThen by r_{Threshold value}It is assigned to and meets regX₂In region unit distance correlation radiusIfThen willIt is assigned to and meets regX₂In region unit distance correlation radius

Enumerate, ifThen by r_{Threshold value}It is assigned to and meets regX₂In region unit distance correlation radiusIfThen willIt is assigned to and meets regX₂In region unit distance correlation radius

Step 203-3：Calculating meets regX₂In any two geographical labels position central point distance；

WithCentral point distance be designated as

Step 203-4：Definition meets regX₂In distance correlation

Enumerate,WithDistance correlation be designated as

In the present invention, statistics is in second region unit regX₂In distance correlation be designated as

Step 204：Semantic dependency

Enumerate, in second region unit regX₂In hospital B string length be designated asLibrary C character string Length is designated as

Enumerate, in second region unit regX₂Institute of traditional Chinese medicine B and library C editing distance is

Enumerate, in second region unit regX₂In hospital B and library C semantic dependency be designated as

In the present invention, in area information Q_31-32In the semantic dependencies of geographical labels be designated as

Step 205：Whether region unit merges

Distance correlation threshold value is set to be designated as rel_Distance, semantic dependency threshold value be designated as rel_{It is semantic}, and according to rel_DistanceWith rel_{It is semantic}Whether the merging treatment of region unit is carried out；

Step 205-1：IfAndWhen, willIt is merged intoIn, thenIt is updated toAnd willExport and give LBSN numbers According to storehouse, step 205-2 is performed；

Step 205-2：IfOrWhen, thenWith Without region merged block, and willWithExport and give LBSN databases.

In the present invention, by second region unit regX₂Write-in LBSN databases 2 register dot position information using collection Conjunction form is expressed as

(3) the 3rd region units

In the present invention, to last region unit regX_yCarry out geographical labels and carry out geographical labels cluster strategy POI- CP process step is：

Step 301：Extract the geographical labels belonged in the same area block

From Q_31-32={ regX₁,regX₂..., regX_yIn extract and meet last region unit regX_yGeographical mark Label, if last region unit regX_yIn geographical labels be only D, then meet regX_yGeographical labels be；

Step 302：The position number of geographical labels is obtained；

To last region unit regX_yIn geographical labels D carry out position classification, be met regX_yIn D it is geographical Label positionThe D is in regX_yMiddle appearance Number of times is designated asAnd

Step 303：Distance correlation

Step 303-1：ParsingIn warp Spend average valueWith latitude average valueI represents asking in summation relation And index,It is met regX_yIn D geographical labels position place-centric point ParsingIn each label point position arrive Distance, and select maximum linear distance, be designated as

In the present invention, statistics is in last region unit regX_yIn all geographical labels and geographical labels position Place-centric pointBetween maximum linear distanceAnd

Step 303-2：Setting area block radius threshold is designated as r_{Threshold value}；

If maximum linear distance is less than zone radius threshold value r_{Threshold value}, then by r_{Threshold value}It is assigned to the region unit distance in affiliated area Correlation radius rD_POI；

If maximum linear distance is more than or equal to zone radius threshold value r_{Threshold value}, then choose maximum linear distance and be used as affiliated area In region unit distance correlation radius rD_POI；

Enumerate, ifThen by r_{Threshold value}It is assigned to and meets regX_yIn region unit distance correlation radiusIfThen willIt is assigned to and meets regX_yIn region unit distance correlation radius

Step 303-3：Calculating meets regX_yIn any two geographical labels position central point distance；

Due to meeting regX_yIn geographical labels there was only D, therefore r_{Threshold value}Centered on point distance be designated as

Step 303-4：Definition meets regX_yIn distance correlation

Enumerate, willDistance correlation be designated as

Step 304：Semantic dependency

Enumerate, in last region unit regX_yIn teaching building D string length be designated as

Enumerate, in last region unit regX_yIn due to there was only teaching building D, therefore D editing distance is designated asAnd

Enumerate, in last region unit regX_yIn teaching building D semantic dependency be designated as And

Step 305：Whether region unit merges

Distance correlation threshold value is set to be designated as rel_Distance, semantic dependency threshold value be designated as rel_{It is semantic}, and according to rel_DistanceWith rel_{It is semantic}Whether the merging treatment of region unit is carried out；

In the present invention, due toWithThereforeWith Therefore last region unit regX_yRegion merging technique need not be carried out.

In the present invention, by last region unit regX_yWrite-in LBSN databases 2 register dot position information use Aggregate form is expressed as

Hot spot region event computing module 33

The first aspect of hot spot region event computing module 33 receives the hot spot region inquiry request Request of user 1, described Request={ Geo (x, y), dist, Hot }, and Request={ Geo (x, y), dist, Hot } is transmitted to LBSN databases 2；

The second aspect of hot spot region event computing module 33 can be according to Request={ Geo (x, y), dist, Hot } The hot spot region matched with Geo (x, y) is searched out in LBSN databases 2, inquiry return information Q is designated as_2-33；

The third aspect of hot spot region event computing module 33 is according to the frequency strategy POI-TP that registers under time window to institute State Q_2-33Calculating processing is carried out, region focus incident and ranking ChecFreq is obtained, and the ChecFreq is fed back into user 1。

In the present invention, hot spot region event computing module 33 can calculate the focus that user 1 asks scope inner region in real time Event, is specifically to ask in respective region that (i.e. current time t's) goes through the moment by the end of request to hot spot region cluster data storehouse 2 History, which contains geography, registers record

In the present invention, with any one region regX_yRegistered according to historical geography exemplified by label detects focus incident, it is fixed Adopted time window is Δ T, and the frequency of registering occurred in Δ T time section is ChecFreq.The frequency of registering ChecFreq refers to the number of times that identical hot spot region is registered in Δ T.The Δ T=| t- (t-1) |, current time is t, previous Moment is t-1.

In the present invention, event is defined as any one region regX_yIn register frequency ChecFreq variable quantity, then event Description expression-form is Trend：

When Trend exceeds certain event detection threshold value Trend_{Threshold value}When, hot spot region event computing module 33 is by region regX_yLabeled as focus incident.And according to the positive and negative event flag is surge type or drops type suddenly of Trend values.

In the present invention, focus incident is ranked, i.e. the more rankings of focus incident time window number are preceding. Any one regX_yIn, it there are under continuous time window Δ T, if the frequency ChecFreq that registers of focus incident exceedes label To frequency threshold value ChecFreq_{Threshold value}, then the continuous duration window number of focus incident more than threshold value is chosen, is designated asSimilarly understand, in all hot spot regions, the continuous duration window number of focus incident is designated as

The active degree Rank and Δ T of event register frequency and lasting event of registering is directly proportional, i.e.,：

WhereinRepresent the quantity of registering in time window Δ T, regX_yAny one region is represented, that is, is summed Element, i represents summing target, Ω_iRepresent that user 1 asks scope inner region to be registered total quantity in time window Δ T.Represent the time window number in current all hot spot regionsMaximum, j represents time window The maximum area identification number of number.

A kind of hot spot region incident detection system registered based on geographical labels proposed by the present invention, the system, which belongs to, to be based on Event detection technology field in the social network of position.System uninterrupted always can must run LBSN and crawl module, warp-wise first The record of registering containing geographical labels is write in LBSN databases, next general area can be obtained using clustering algorithm of registering Cluster；Then the region clustering module of system can calculate precise region using the region clustering algorithm containing geographical labels；Heat The request of point area calculation module meeting relative users, this is the service interface that system is uniquely externally provided, and the module can root first The query argument submitted according to user, submits corresponding inquiry, and carry out region focus incident to returning to the data come to database Detection algorithm, calculates the ranking of focus incident and event, and returns to request user.

Embodiment

Referring to shown in Fig. 1, Fig. 2, it is assumed that the user that geographical labels are met in LBSN databases registers information Q_31-2=R_ POI_p(x, y), POI search result has multiple, then application cluster interval time kcluster-span progress k-means clusters Method processing, obtains region block message Q_31-32={ regX₁,regX₂,…,regX_y}.Described Q_31-2=R_POI_p(x,y), POI is original record, is not clustered also.

Assuming that Q_31-32={ regX₁,regX₂,…,regX_yCoarseness geographic area after cluster is 6, then region block number Y=6, i.e. Q_31-32={ regX₁,regX₂,regX₃,regX₄,regX₅,regX₆}。

In Q_31-32Interior, interregional distance relation such as following table：

Assuming that when request user sends local hot spot region inquiry request, the running of system is as follows：

Step 1：The recommendation request information for asking user is Request={ Geo (x, y), dist, Hot }={ inR₃, 2000, Hot }, inR₃For the location point of request, it is with other regional distances：

It can be seen that, in the range of request, there are 3 regions to meet condition, i.e. regX₁, regX₃, regX₄.The request of user It can be received first by the hot spot region event computing module 33 of hot spot region incident detection system.

Step 2：Hot spot region event computing module 33 comes into user the parametric configuration come please into rational query statement Database is sought, database will then meet desired these three regions regX₁, regX₃, regX₄With they with geographical labels History is registered recordReturn to hot spot region event and calculate mould Block 33.

Step 3：Hot spot region event computing module 33 obtains registering after record, proceeds by calculating.Δ T value is 1 small When, frequency of registering threshold value ChecFreq_{Threshold value}For 100, event detection threshold value Trend_{Threshold value}For 50/h, 4 times are found forward Window, first time window T₁=t-3 Δs T, second time window T₂=t-2 Δ T, the 3rd time window T₃=t- Δs T, the 4th time window T₄=t,.

In region regX₁Interior, the frequency ChecFreq that registers of a certain event is：

Period	T1	T2	T3	T4
					ChecFreq	167	101	150	50

Its continuous active time window value of maximum (the i.e. continuous duration window number of focus incident that is 3)。

In region regX₃Interior, the frequency ChecFreq that registers of a certain event is：

Period	T1	T2	T3	T4
					ChecFreq	24	30	50	99

Its continuous active time window value of maximum is 3.

In region regX₄Interior, the frequency ChecFreq that registers of present period is：

Period	T1	T2	T3	T4
					ChecFreq	112	22	23	12

Its continuous active time window value of maximum is 1.

According toEvent detection threshold value is exceeded； Not up to event detection threshold value；Event detection threshold value is exceeded.It can be seen that there is two hot spot regions.Point It is not regX₁And regX₄。

For regX₁, its continuous active time window value is 3, the active degree ranking of its event：

For regX₄, its continuous active time window value is 1, the active degree ranking of its event：

Result of calculation is returned to user by hot spot region event computing module 33.That is ChecFreq returning results are ChecFreq={ regX₁:1,regX₄:2}。

Claims (1)

1. a kind of hot spot region incident detection system based on geographical labels applied to LBSN networks, it is characterised in that：It is described The hot spot region incident detection based on geographical labels is set between the LBSN databases (2) and user (1) in LBSN networks System (3)；

The hot spot region incident detection system (3) based on geographical labels includes cluster module of registering (31), based on label The area calculation module (32) and hot spot region event computing module (33) of cluster；The hot spot region event computing module (33) To be connected interface between LBSN databases (2) and user (1)；

Cluster module (31) first aspect of registering is used to send the solicited message of registering containing geographical labels to LBSN databases (2) Q_31-2, the Q_31-2=R_POI_p(x,y),POI；

R_POI_p(x, y) represents sign-in desk geographical position, and x represents longitude, and y represents latitude；

POI represents geographical labels；Any one geographical labels in the POI are designated as a, and another geographical labels is designated as b, a, b ∈POI；

Cluster module (31) second aspect of registering is according to Q_31-2=R_POI_p(x, y), POI can be searched in LBSN databases (2) Go out the record of registering matched with geographical labels POI, be designated as the return information Q that registers_2-31；

Cluster module (31) third aspect of registering is to the return information Q that registers that receives_2-31According to cluster interval time Kcluster-span carries out k-means clustering method processing, obtains region block message Q_31-32, the Q_31-32={ regX₁, regX₂,…,regX_y′, then by Q_31-32Export to the area calculation module (32) based on label clustering；

regX₁Represent first region unit in the R of any one geographic area；

regX₂Represent second region unit in the R of any one geographic area；

regX_y′Represent last region unit in the R of any one geographic area；

Y ' expression region unit numbers；

Area calculation module (32) first aspect based on label clustering is used for receiving area block message Q_31-32={ regX₁, regX₂,…,regX_y′}；

Area calculation module (32) second aspect based on label clustering clusters strategy POI-CP to Q according to geographical labels_31-32= {regX₁,regX₂,…,regX_y′Handled, obtain convergence geographical labels region unit Q_32-2；And will convergence geographical labels region Block is written to LBSN databases (2)；

Hot spot region event computing module (33) first aspect receives the hot spot region inquiry request Request of user (1), described Request={ Geo (x, y), dist, Hot }, and Request={ Geo (x, y), dist, Hot } is transmitted to LBSN databases (2)；

Geo (x, y) in the hot spot region inquiry request Request={ Geo (x, y), dist, Hot } represents request user The geographical position at place, x is longitude, and y is latitude；Dist represents the interest distance radius that user is set；Hot represents that user pays close attention to Hot spot region；

Hot spot region event computing module (33) second aspect foundation hot spot region inquiry request Request=Geo (x, y), Dist, Hot } hot spot region matched with Geo (x, y) can be searched out in LBSN databases (2), it is designated as inquiring about return information Q_2-33；

Hot spot region event computing module (33) third aspect is according to the frequency strategy POI-TP that registers under time window to described Q_2-33Calculating processing is carried out, region focus incident and ranking ChecFreq is obtained, and the ChecFreq is fed back into user (1)；

Described geographical labels cluster strategy POI-CP has the following steps：

Extraction belongs to the same area block regX_y′In geographical labels POI the step of；

Calculating belongs to the same area block regX_y′In geographical labels POI position numberThe step of；

Calculate geographical labels POI and the place-centric point of geographical labels positionBetween Maximum linear distanceThen judge describedWith zone radius threshold value r_{Threshold value}Size, ifThen by r_{Threshold value}It is assigned to the region unit distance correlation radius in affiliated areaIfThen choose maximum linear distance and be used as the region unit distance correlation radius in affiliated areaAnd then the distance correlation radius for passing through a geographical labelsWith the distance correlation radius of b geographical labelsSum is than the central point distance in upper geographical labels POI between any two geographical labels a, bObtain Distance correlationThe step of；For longitude average value,For Latitude average value；

Calculate the semantic dependency between any two geographical labels a, b in geographical labels POI The step of；For editing distance,For string length；

According to describedWith it is describedWith distance correlation threshold value rel_Distance, semantic dependency threshold Value rel_{It is semantic}Contrasted, and according to comparing result combined region block regX_y' the step of；

IfAndWhen, by the position of registering of b geographical labels It is merged into the position of registering of a geographical labels

IfOrWhen, then the position of registering of b geographical labelsNo With the position of registering of a geographical labelsCarry out region merged block；

The concrete mode of the frequency strategy POI-TP that registers under described time window is：Hot spot region event computing module (33) User (1) can be calculated in real time ask any one region regX_y′Focus incident, when to hot spot region cluster data storehouse (2) ask History in respective region by the end of current time t contains geography and registered and records Q_2-33, then event description beΔ T represents time window, Δ T=| t- (t-1) |, when t represents current Carve, t-1 represents previous moment, ChecFreq^tRepresent the frequency of registering at current time t, ChecFreq^t-1Represent previous moment t-1 Frequency of registering；

The active degree Rank and Δ T of event register frequency and lasting event of registering is directly proportional, i.e.,：

$Rank=\underset{i=1}{\overset{{\mathrm{regX}}_{y^{\′}}}{\Σ}}\frac{{\mathrm{ChecFreq}}_i^t}{{\mathrm{\Ω}}_i}\×\sqrt{1+\frac{\underset{j\∈{\mathrm{regX}}_{y^{\′}}}{max}\left({\mathrm{CU}}^j\right)}{{\mathrm{regX}}_{y^{\′}}}};$

WhereinRepresent the frequency of registering in time window Δ T, regX_y′Any one region is represented, that is, member of summing Element, i represents summing target, Ω_iRepresent that user (1) asks scope inner region to be registered total quantity in time window Δ T,Represent the time window number in current all hot spot regionsMaximum, j represents time window The maximum area identification number of mouth number.