CN116340600A - Push method of network appointment vehicle on-vehicle location, training method and device of prediction model - Google Patents
Push method of network appointment vehicle on-vehicle location, training method and device of prediction model Download PDFInfo
- Publication number
- CN116340600A CN116340600A CN202310150283.7A CN202310150283A CN116340600A CN 116340600 A CN116340600 A CN 116340600A CN 202310150283 A CN202310150283 A CN 202310150283A CN 116340600 A CN116340600 A CN 116340600A
- Authority
- CN
- China
- Prior art keywords
- user
- location
- taxi
- boarding
- sample
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 74
- 238000012549 training Methods 0.000 title claims abstract description 60
- 230000001960 triggered effect Effects 0.000 claims abstract description 10
- 238000000605 extraction Methods 0.000 claims description 90
- 238000012545 processing Methods 0.000 claims description 36
- 230000015654 memory Effects 0.000 claims description 27
- 230000006399 behavior Effects 0.000 claims description 24
- 238000013136 deep learning model Methods 0.000 claims description 22
- 238000012216 screening Methods 0.000 claims description 10
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 230000006978 adaptation Effects 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 7
- 238000012360 testing method Methods 0.000 description 5
- 238000012795 verification Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000002085 persistent effect Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000000452 restraining effect Effects 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/907—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/909—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using geographical or spatial information, e.g. location
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/903—Querying
- G06F16/9035—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/903—Querying
- G06F16/9038—Presentation of query results
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/40—Business processes related to the transportation industry
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Library & Information Science (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a pushing method of a network taxi on-board place, a training method and a training device of a prediction model. The pushing method of the network appointment vehicle on-vehicle places comprises the following steps: responding to a vehicle-restraining request triggered by a user, and acquiring vehicle-restraining habit information of the user; determining a vehicle-restraining place when a user triggers a vehicle-restraining request; at least one candidate boarding location corresponding to the recall location, and determining location attribute information of each candidate boarding location; determining an optional boarding location to be recommended from at least one candidate boarding location based on the taxi-taking habit information and location attribute information of each candidate boarding location; the selectable boarding location is pushed to the user for the user to determine a target boarding location based on the selectable boarding location. By the method and the device, the adaptation degree between the target boarding location and the user can be improved, so that the user can quickly reach the target boarding location, and the taxi-restraining experience of the user is improved.
Description
Technical Field
The invention relates to the field of cloud computing, in particular to a pushing method of a network about vehicle on-vehicle place, a training method of a prediction model and a device thereof.
Background
With the rapid development of network taxi taking, the taxi taking mode of people is changed from hand-in to stop to online reservation, and further, the taxi taking method is greatly convenient for people to go out. But the driver and the passengers often cannot find each other due to inaccurate boarding places by adopting an online reservation mode to drive, so that the journey of the passengers is delayed, and even the disputes of driving and riding are generated.
In the related art, an Application (APP) recommends a boarding location with the highest frequency to a user according to the frequency of use of each boarding location in an area where the user is located when the user places a order, so that a driver can quickly arrive at the boarding location. However, the method is adopted to recommend the boarding location, and the preference of the user and the influence of the boarding roll names in the surrounding area on the boarding are not considered, so that the user cannot find the correct boarding location, and further the car restraining experience of the user is influenced.
Disclosure of Invention
Therefore, the technical problem to be solved by the invention is to overcome the defect that the get-on place recommended for the user for the network taxi company in the prior art is not targeted and influences the taxi-taking experience of the user, so as to provide a pushing method of the network taxi-on place, a training method of a prediction model and a device thereof.
According to a first aspect, an embodiment of the present invention provides a pushing method for a network about a vehicle-on-vehicle location, where the method includes:
responding to a vehicle-restraining request triggered by a user, and acquiring vehicle-restraining habit information of the user;
determining a vehicle-restraining place when the user triggers the vehicle-restraining request;
recalling at least one candidate boarding location corresponding to the taxi-taking location, and determining location attribute information of each candidate boarding location;
determining an optional boarding location to be recommended from the at least one candidate boarding location based on the taxi-habit information and location attribute information of each candidate boarding location;
pushing the optional boarding location to the user for the user to determine a target boarding location based on the optional boarding location.
In the mode, the optional boarding location conforming to the user taxi-taking habit can be screened out of at least one candidate boarding location corresponding to the taxi-taking location for the user based on the taxi-taking habit information of the user and the location attribute information of the taxi-taking location, and the determined optional boarding location is recommended to the user, so that the user can determine the target boarding location from the recommended optional boarding locations, the adaptation degree between the target boarding location and the user is improved, the user can quickly reach the target boarding location, and the taxi-taking experience of the user is improved.
With reference to the first aspect, in a first embodiment of the first aspect, the determining, based on the about habit information and the location attribute information of each candidate boarding location, an optional boarding location to be recommended from the at least one candidate boarding location includes:
respectively carrying out feature extraction processing on the vehicle habit information and the attribute information of each place to obtain a feature extraction result;
inputting the feature extraction result into a pre-trained recommendation prediction model, and respectively determining recommendation confidence of each candidate boarding location to be recommended to the user;
and determining optional boarding places to be recommended from the at least one candidate boarding place based on recommendation confidence of each candidate boarding place to be recommended to the user.
With reference to the first embodiment of the first aspect, in a second embodiment of the first aspect, the determining, based on a recommendation confidence level of each candidate boarding location to be recommended to the user, an optional boarding location to be recommended from the at least one candidate boarding location includes:
and sequencing the candidate boarding places according to the order of the recommendation confidence coefficient of each candidate boarding place to be recommended to the user from high to low, and selecting N candidate boarding places as selectable boarding places to be recommended according to the order of the recommendation confidence coefficient from high to low, wherein N is an integer greater than 0.
With reference to the first embodiment of the first aspect, in a third embodiment of the first aspect, the determining, based on a recommendation confidence level of each candidate boarding location to be recommended to the user, an optional boarding location to be recommended from the at least one candidate boarding location includes:
and determining the candidate boarding places with the recommendation confidence coefficient larger than or equal to the confidence coefficient threshold value as the selectable boarding places to be recommended according to the recommendation confidence coefficient of each candidate boarding place to be recommended to the user.
With reference to the first aspect, the first embodiment of the first aspect, the second embodiment of the first aspect, or the third embodiment of the first aspect, in a fourth embodiment of the first aspect, the performing feature extraction processing on the vehicle habit information and each piece of location attribute information to obtain feature extraction results includes:
performing feature extraction processing on the vehicle-restraining habit information to obtain vehicle-restraining behavior features of the user and vehicle-restraining time features of the user;
respectively carrying out feature extraction processing on each place attribute information to obtain the heat feature of each place attribute information corresponding to the candidate boarding place and the geographic feature of each place attribute information corresponding to the candidate boarding place;
And taking the feature of the taxi-taking behavior of the user, the feature of the taxi-taking time of the user, the heat feature of each place attribute information corresponding to the candidate taxi-taking place and the geographic feature of each place attribute information corresponding to the candidate taxi-taking place as feature extraction results.
With reference to the second embodiment or the third embodiment of the first aspect, in a fifth embodiment of the first aspect, the pushing the optional boarding location to the user includes:
and pushing the optional boarding places to the user according to the sequence of the recommendation confidence from high to low.
With reference to the first aspect, in a sixth embodiment of the first aspect, the recalling at least one candidate boarding location corresponding to the about location includes:
determining the area of the vehicle-restraining place;
and recalling all boarding places in the area as at least one candidate boarding place corresponding to the taxi-taking place.
With reference to the sixth embodiment of the first aspect, in a seventh embodiment of the first aspect, the method further includes:
if no boarding location exists in the area, determining a road where the taxi-taking location is located;
and taking the boarding location corresponding to the road as a candidate boarding location and recalling.
According to a second aspect, an embodiment of the present invention further provides a training method of a prediction model, where the training method of the prediction model includes:
acquiring historical taxi-taking information of a plurality of user samples within a specified time range, wherein the historical taxi-taking information comprises taxi-taking habit sample information corresponding to the user samples and place sample attribute information of a target taxi-taking place selected by each taxi-taking;
respectively carrying out feature extraction processing on the vehicle habit sample information of each user sample and the corresponding place sample attribute information to obtain a plurality of feature sample extraction results;
training a deep learning model based on the extraction results of the plurality of feature samples to obtain a recommendation prediction model, wherein the recommendation prediction model is used for predicting recommendation confidence of a candidate on-board location to be recommended to a user.
In the mode, based on the historical taxi-taking information of a plurality of user samples, the selection relation between the taxi-taking habits of the users and the target taxi-taking places corresponding to different taxi-taking places can be fully excavated, and then the dependence on experience of selecting the taxi-taking places can be avoided when the selectable taxi-taking places are recommended for the users later, so that the recommendation effectiveness of the selectable taxi-taking places is improved.
With reference to the second aspect, in a first embodiment of the second aspect, the performing feature extraction processing on the about habit sample information and the corresponding location sample attribute information of each user sample to obtain a plurality of feature sample extraction results includes:
feature extraction processing is carried out on the taxi-taking habit sample information of the current user sample, and a taxi-taking behavior feature sample of the current user sample and a taxi-taking time feature sample of the current user sample are obtained;
respectively carrying out feature extraction processing on the attribute information of the plurality of place samples of the current user sample to obtain a heat feature sample and a geographic feature sample of each target boarding place corresponding to the current user sample;
and taking the taxi-taking behavior feature sample of the current user sample, the taxi-taking time feature sample of the current user sample, the heat feature sample of each target boarding place corresponding to the current user sample and the geographic feature sample as a plurality of feature sample extraction results of the current user sample.
According to a third aspect, an embodiment of the present invention further provides a pushing device for a network about a vehicle-on-vehicle location, where the device includes:
The first acquisition unit is used for responding to a vehicle-restraining request triggered by a user and acquiring vehicle-restraining habit information of the user;
the first determining unit is used for determining a vehicle-restraining place when the user triggers the vehicle-restraining request;
a recall unit, configured to recall at least one candidate boarding location corresponding to the taxi-taking location, and determine location attribute information of each candidate boarding location;
a screening unit, configured to determine an optional boarding location to be recommended from the at least one candidate boarding location based on the taxi-taking habit information and location attribute information of each candidate boarding location;
and the pushing unit is used for pushing the optional boarding location to the user so that the user can determine a target boarding location based on the optional boarding location.
According to a fourth aspect, an embodiment of the present invention further provides a training device for a prediction model, where the training device for a prediction model includes:
the second acquisition unit is used for acquiring historical taxi-taking information of a plurality of user samples within a specified time range, wherein the historical taxi-taking information comprises taxi-taking habit sample information corresponding to the user samples and place sample attribute information of a target taxi-taking place selected by each taxi-taking;
The second feature extraction unit is used for carrying out feature extraction processing on the about habit sample information of each user sample and the corresponding place sample attribute information respectively to obtain a plurality of feature sample extraction results;
the training unit is used for training the deep learning model based on the extraction results of the plurality of characteristic samples to obtain a recommendation prediction model, and the recommendation prediction model is used for predicting recommendation confidence of the candidate boarding location to be recommended to the user.
According to a fifth aspect, an embodiment of the present invention further provides an electronic device, including a memory and a processor, where the memory and the processor are communicatively connected to each other, and the memory stores computer instructions, and the processor executes the computer instructions, thereby executing the pushing method of the network about a vehicle location of any one of the first aspect and an optional embodiment thereof, or executing the training method of the prediction model of any one of the second aspect and an optional embodiment thereof.
According to a sixth aspect, the embodiment of the present invention further provides a computer readable storage medium, where computer instructions are stored, where the computer instructions are configured to cause the computer to perform the pushing method of the network about a vehicle location according to any one of the first aspect and its alternative embodiments or perform the training method of the prediction model according to any one of the second aspect and its alternative embodiments.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a pushing method of a network about a vehicle-on-vehicle location according to an exemplary embodiment.
Fig. 2 is a flow chart of an alternative method of boarding location determination according to an exemplary embodiment.
Fig. 3 is a schematic diagram illustrating a process of pushing a network about a vehicle-on-vehicle location according to an exemplary embodiment.
FIG. 4 is a flowchart of a method for training a predictive model in accordance with an exemplary embodiment.
Fig. 5 is a flowchart of a method for extracting feature sample extraction results according to an exemplary embodiment.
Fig. 6 is a schematic diagram of a model structure of a deep fm model according to an exemplary embodiment.
Fig. 7 is a block diagram of a pushing device for a network about a vehicle location according to an exemplary embodiment.
Fig. 8 is a block diagram of a training apparatus for a predictive model according to an exemplary embodiment.
Fig. 9 is a schematic diagram of a hardware structure of an electronic device according to an exemplary embodiment.
Detailed Description
The following description of the embodiments of the present invention will be made apparent and fully in view of the accompanying drawings, in which some, but not all embodiments of the invention are shown. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the related art, the internet taxi-taking APP recommends the taxi-taking place with the highest use frequency to the user according to the use frequency of each taxi-taking place in the area where the user is when the user places a order, so that a driver can quickly reach the taxi-taking place. However, the method is adopted to recommend the boarding location, and the preference of the user and the influence of the boarding roll names in the surrounding area on the boarding are not considered, so that the user cannot find the correct boarding location, and further the car restraining experience of the user is influenced.
In order to solve the above-mentioned problems, in the embodiments of the present invention, a pushing method for a network bus-on-bus location is provided for an electronic device, and it should be noted that an execution body of the pushing method may be a pushing device for the network bus-on-bus location, and the pushing device may be implemented as part or all of the electronic device by software, hardware or a combination of software and hardware, where the electronic device may be a terminal, a client, or a server, and the server may be a server, or may be a server cluster formed by multiple servers. In the following method embodiments, the execution subject is an electronic device.
The pushing method of the network vehicle-on-vehicle place is applied to a use scene of a user for vehicle-on-vehicle through the network vehicle-on-vehicle APP. The pushing method of the network about vehicle on-vehicle place provided by the invention comprises the following steps: responding to a vehicle-restraining request triggered by a user, and acquiring vehicle-restraining habit information of the user; determining a vehicle-restraining place when a user triggers a vehicle-restraining request; at least one candidate boarding location corresponding to the recall location, and determining location attribute information of each candidate boarding location; determining an optional boarding location to be recommended from at least one candidate boarding location based on the taxi-taking habit information and location attribute information of each candidate boarding location; the selectable boarding location is pushed to the user for the user to determine a target boarding location based on the selectable boarding location. According to the pushing method for the network taxi-taking place, which is provided by the invention, the selectable taxi-taking place which accords with the taxi-taking habit of the user can be screened out of at least one candidate taxi-taking place corresponding to the taxi-taking place based on the taxi-taking habit information of the user and the place attribute information of the taxi-taking place, and the determined selectable taxi-taking place is recommended to the user, so that the user can determine the target taxi-taking place from the recommended selectable taxi-taking places, the adaptation degree between the target taxi-taking place and the user is improved, the user can quickly reach the target taxi-taking place, and the taxi-taking experience of the user is improved.
Fig. 1 is a flowchart of a pushing method of a network about a vehicle-on-vehicle location according to an exemplary embodiment. As shown in fig. 1, the pushing method of the network about on-board location includes the following steps S101 to S105.
In step S101, in response to receiving a vehicle-restraining request triggered by a user, vehicle-restraining habit information of the user is acquired.
In the embodiment of the invention, a user can trigger a vehicle restraint request through the network vehicle restraint APP of the client. And responding to the vehicle-restraining request triggered by the user, determining that the user needs to request the online vehicle-restraining, and further acquiring the vehicle-restraining habit information of the user so as to know the vehicle-restraining habit of the user through the vehicle-restraining habit information of the user, so that the vehicle-restraining requirement of the user can be met with more pertinence when the user recommends the optional vehicle-restraining place for the user later.
In step S102, a vehicle-restraining place when the user triggers a vehicle-restraining request is determined.
In the embodiment of the invention, the vehicle-restraining place can be understood as a real-time position when a user triggers a vehicle-restraining request. For example: the taxi-taking place may be a park, office building or scenic spot where the user triggers a taxi-taking request, etc.
In step S103, at least one candidate boarding location corresponding to the appointment location is recalled, and location attribute information of each candidate boarding location is determined.
In the embodiment of the invention, the candidate boarding location can be understood as a boarding location near the boarding location, where a user can sit on a net to board. After at least one candidate boarding location corresponding to the taxi-taking location is recalled, location attribute information of each recalled candidate boarding location is determined, so that the selected heat and the geographic environment of the corresponding candidate boarding location are determined through the location attribute information, and further recommendation effectiveness of the selectable boarding location is improved.
In one embodiment, the candidate pick-up location may be determined based on the area in which the pick-up location is located. Specifically, determining the area of the taxi-taking place, recalling all taxi-taking places in the area as at least one candidate taxi-taking place corresponding to the taxi-taking place, and further helping to avoid the fact that the recalled candidate taxi-taking place is far away from the taxi-taking place and the effectiveness of recommendation is affected. In one example, taking the about car place as an office building in the XX park, the XX park is the area where the about car place is located, and the place where the XX park can be used for the user to take the network about car is the candidate boarding place corresponding to the about car place. For example, the location where the XX park is available for the user to travel on the net may be the gate of the XX park in a certain direction.
In another embodiment, if no boarding location exists in the area, determining the road where the boarding location is located, and further taking the boarding location corresponding to the road as a candidate boarding location and recalling the candidate boarding location, so as to ensure that the boarding location can be recommended for the user, and meeting the boarding requirement of the user.
In yet another embodiment, the area or road where the about vehicle location is located may be determined by a GPS or road map.
In step S104, an optional boarding location to be recommended is determined from at least one candidate boarding location based on the appointment habit information and location attribute information of each candidate boarding location.
In the embodiment of the invention, the optional boarding location can be understood as the boarding location which is relatively more in line with the taxi-taking habit of the user and can meet the taxi-taking requirement of the user in all the recalled candidate boarding locations.
In step S105, the selectable boarding location is pushed to the user for the user to determine a target boarding location based on the selectable boarding location.
According to the embodiment, the optional boarding location which accords with the user boarding habit can be screened from at least one candidate boarding location corresponding to the boarding location for the user based on the user boarding habit information and the location attribute information of the boarding location, and the determined optional boarding location is recommended to the user, so that the user can determine the target boarding location from the recommended optional boarding locations, the adaptation degree between the target boarding location and the user is improved, the user can quickly reach the target boarding location, and the boarding experience of the user is improved.
The following examples will specifically describe a process of determining an alternative boarding location to be recommended from at least one candidate boarding location. As shown in fig. 2, the determination of the optional boarding location includes the following steps.
In step S201, feature extraction processing is performed on the vehicle habit information and the location attribute information, respectively, to obtain feature extraction results.
In the embodiment of the invention, since the vehicle-engaging habit information and the place attribute information belong to text information, in order to facilitate the prediction of a pre-trained recommended prediction model, the vehicle-engaging habit information and the place attribute information are respectively subjected to feature extraction processing in a coding mode, so that a feature extraction result is obtained. In one example, the encoding may be one-hot encoding (one-hot) or compression encoding, and may be set according to actual requirements, and is not limited in the present invention.
Specifically, feature extraction processing is performed on the taxi-taking habit information to obtain taxi-taking behavior features of the user and taxi-taking time features of the user. Wherein the vehicle behavior feature may include, but is not limited to, the following sub-features: historical taxi-taking places, walking distances between the historical taxi-taking places and corresponding target taxi-taking places, and whether the user manually determines the target taxi-taking places before taxi-taking each time. Wherein the act of manually determining whether the target boarding location exists comprises: whether there is a search behavior to search for a target boarding location, and whether there is a dragging behavior to overadjust the target boarding location. The time of appointment characteristics may include, but are not limited to, the following: the user saves the time interval between the time of the holiday car, the time of the work day car and the last time of the holiday car. By extracting the vehicle-restraining behavior characteristics and the vehicle-restraining time characteristics of the user, the vehicle-restraining habit of the user in the history of vehicle-restraining can be clarified.
And respectively carrying out feature extraction processing on each place attribute information to obtain the heat feature of each place attribute information corresponding to the candidate boarding place and the geographic feature of each place attribute information corresponding to the candidate boarding place. Wherein the heat signature includes, but is not limited to, the following sub-signatures: the heat characteristic of the target get-on place, the order heat characteristic of the about place and the corresponding about order completion heat characteristic. Geographic features include, but are not limited to, the following sub-features: the road grade of the candidate get-on place, the road width, the road condition of the car during the car-taking, and whether the road of the candidate get-on place is forbidden or not. By extracting the heat characteristics and the geographic characteristics of the candidate boarding places, the history selection condition of each candidate boarding position can be clarified, and the prediction effectiveness can be improved when the recommendation prediction is carried out subsequently.
Taking the characteristic of the taxi-taking behavior of the user, the characteristic of the taxi-taking time of the user, the heat characteristic of the candidate taxi-taking place corresponding to each place attribute information and the geographic characteristic of the candidate taxi-taking place corresponding to each place attribute information as characteristic extraction results, so that the possibility that each candidate taxi-taking place can be selected by the user can be predicted through a pre-trained recommendation prediction model, and further the pushing effectiveness of the selectable taxi-taking place can be improved.
In one example, if there is a sub-feature for which a specific feature is not extracted in the process of extracting the user's travel feature, the user's travel time feature, the heat feature of each location attribute information corresponding to the candidate boarding location, and the geographic feature of each location attribute information corresponding to the candidate boarding location, a default feature corresponding to the sub-feature is adopted as the feature extraction result of the sub-feature.
In step S202, the feature extraction result is input to a pre-trained recommendation prediction model, and recommendation confidence of each candidate boarding location to be recommended to the user is determined respectively.
In step S203, an optional boarding location to be recommended is determined from at least one candidate boarding location based on the recommendation confidence level to be recommended to the user for each candidate boarding location.
In the embodiment of the invention, the number of the candidate boarding places can be multiple or one. Therefore, in order to improve the pushing effectiveness and shorten the selection time of determining the target boarding location by the user, the selectable boarding location to be recommended is determined from at least one candidate boarding location based on the recommendation confidence coefficient of each candidate boarding location to be recommended to the user, so that the taxi-taking efficiency of the user is improved.
In an embodiment, the candidate boarding places are ordered according to the order of the recommendation confidence coefficient of each candidate boarding place to be recommended to the user from high to low, and N candidate boarding places are selected as optional boarding places to be recommended according to the order of the recommendation confidence coefficient from high to low, so that when the optional boarding places are pushed subsequently, the optional boarding places can be pushed according to the order of the recommendation confidence coefficient, and the user can quickly obtain the optimal target boarding places. Wherein N is an integer greater than 0, and specific values may be set according to actual requirements, which is not limited in the present invention. For example: if there are 5 candidate get-on places, namely a get-on place A, a get-on place B, a get-on place C, a get-on place D and a get-on place E, wherein the recommendation confidence corresponding to the get-on place A is 0.95, the recommendation confidence corresponding to the get-on place B is 0.65, the recommendation confidence corresponding to the get-on place C is 0.78, the recommendation confidence corresponding to the get-on place D is 0.94 and the recommendation confidence corresponding to the get-on place E is 0.55, and the candidate get-on places are ordered according to the order of the recommendation confidence from high to low to obtain the following ordering result: a boarding location A, a boarding location D, a boarding location C, a boarding location B and a boarding location E. If N is 3, taking the boarding location A, the boarding location C and the boarding location D as optional boarding locations to be recommended.
In another embodiment, according to the recommendation confidence coefficient of each candidate boarding location to be recommended to the user, the candidate boarding location with the recommendation confidence coefficient greater than or equal to the confidence coefficient threshold value is determined as the optional boarding location to be recommended, so that the pushing efficiency is improved. For example: if there are 5 candidate get-on places, namely a get-on place A, a get-on place B, a get-on place C, a get-on place D and a get-on place E, wherein the recommendation confidence corresponding to the get-on place A is 0.95, the recommendation confidence corresponding to the get-on place B is 0.65, the recommendation confidence corresponding to the get-on place C is 0.75, the recommendation confidence corresponding to the get-on place D is 0.94 and the recommendation confidence corresponding to the get-on place E is 0.55, and the confidence threshold is 80%, the get-on place A and the get-on place D are taken as optional get-on places to be recommended.
Through the embodiment, the recommended optional boarding places are more targeted and more accord with the taxi-taking habit of the user, so that the taxi-taking experience of the user is improved.
In an embodiment, when the selectable boarding location is pushed to the user, the selectable boarding location with the highest recommendation confidence level can be pushed to the user according to the sequence from high to low of the recommendation confidence level, so that the user can be helped to prioritize the selectable boarding location with the highest recommendation confidence level when determining the target boarding location.
In an implementation scenario, the pushing process of the network about the on-board location may be as shown in fig. 3. And responding to the received vehicle-restraining request triggered by the user, acquiring vehicle-restraining habit information of the user, and determining a vehicle-restraining place when the user triggers the vehicle-restraining request. And recalling at least one candidate boarding location corresponding to the taxi-taking location, and determining location attribute information of each candidate boarding location. The candidate boarding places can be all boarding places in the area where the passing destination places are located, and can be boarding places corresponding to roads where the destination places are located. Based on the vehicle-restraining habit information and the place attribute information of each candidate boarding place, the recommendation confidence of each candidate boarding place to be recommended to the user is respectively determined through a pre-trained recommendation prediction model. N candidate boarding places are selected as optional boarding places to be recommended according to the sequence from high to low of the recommendation confidence, and the optional boarding places are pushed to a user according to the sequence from high to low of the recommendation confidence, so that the user can determine target boarding places based on the optional boarding places, and the success rate of taxi reservation is improved. In one example, the recommended prediction model may be obtained by training a depth factor molecular machine (Deep Factorization Machine, deep fm) model.
In another embodiment, if the number of the candidate boarding places is only one or less than the specified number threshold, the candidate boarding places can be directly used as optional boarding places for pushing, so that pushing efficiency is improved, and a user can quickly get on or off the bus.
Based on the same inventive concept, the invention also provides a training method of the prediction model.
The training method of the prediction model provided by the invention comprises the following steps: acquiring historical vehicle-engaging information of a plurality of user samples within a specified time range, wherein the historical vehicle-engaging information comprises vehicle-engaging habit sample information corresponding to the user samples and place sample attribute information of a target vehicle-engaging place selected by each vehicle-engaging; respectively carrying out feature extraction processing on the vehicle habit sample information of each user sample and the corresponding place sample attribute information to obtain a plurality of feature sample extraction results; training a deep learning model based on the extraction results of the plurality of feature samples to obtain a recommendation prediction model, wherein the recommendation prediction model is used for predicting recommendation confidence of a candidate on-board location to be recommended to a user. According to the training method of the prediction model, provided by the invention, the selection relation between the user appointment habit and the target boarding location can be fully excavated based on the historical taxi appointment information of a plurality of user samples, and then the dependence on the selection experience of the boarding location can be avoided when the selectable boarding location is recommended for the user, so that the recommendation effectiveness of the selectable boarding location is improved.
FIG. 4 is a flowchart of a method for training a predictive model in accordance with an exemplary embodiment. As shown in fig. 4, the training method of the prediction model includes the following steps S401 to S403.
In step S401, historical vehicle information of a plurality of user samples within a specified time range is acquired.
In the embodiment of the invention, in order to improve training efficiency and avoid excessive training, historical vehicle information of a plurality of user samples in a specified time range is acquired. The historical vehicle-restraining information of the user sample is real data information of successful vehicle-restraining of the user and can be obtained through random selection of an online database. The historical taxi-taking information comprises taxi-taking habit sample information corresponding to a user sample and place sample attribute information of a target taxi-taking place selected by each taxi-taking. In one example, the specified time range may be within the last three months before the recommended prediction model is trained, thereby helping to ensure the effectiveness of the training.
In one example, prior to training the deep learning model, historical vehicle information for a plurality of user samples may be partitioned according to random proportions to obtain a training set, a test set, and a validation set for training the deep learning model. Training the deep learning model by using a training set, testing the training progress of the deep learning model by using a testing set, verifying the trained deep learning model by using a verification set, and further determining that the deep learning model is trained to be finished when the verification result meets the specified requirement, so as to obtain the recommended prediction model. Preferably, the division ratio among the training set, the test set and the verification set may be: 5:3:2.
In step S402, feature extraction processing is performed on the vehicle habit sample information and the corresponding location sample attribute information of each user sample, so as to obtain a plurality of feature sample extraction results.
In step S403, a deep learning model is trained based on the plurality of feature sample extraction results, and a recommended prediction model is obtained.
In the embodiment of the invention, the recommendation prediction model is used for predicting the recommendation confidence of the candidate boarding location to be recommended to the user. Because the deep learning model has stronger generalization capability, expansibility and extensibility, more areas can be covered by one model, and then when the deep learning model is trained based on the extraction results of a plurality of feature samples, the selection relation between the user taxi-taking habit and the target taxi-taking places corresponding to different taxi-taking places can be fully excavated, so that the trained recommendation prediction model can be more targeted and effective when predicting the recommendation confidence coefficient of each candidate taxi-taking place selected by the user.
In one example, the deep learning model may be any one of a number of specified network models, such as: the convolutional neural network model, the stack self-coding network model or the deep trust network model can be set according to actual requirements, and the convolutional neural network model, the stack self-coding network model or the deep trust network model is not limited in the invention. Preferably, a depth factor molecular machine (Deep Factorization Machine, deep fm) model may be employed as the deep learning model to be trained. The deep FM model is a deep learning model formed by an FM model and a deep neural network model (Deep Neural Network, DNN), and can learn the interaction relation between low-order features and high-order features, so that the selection relation between the user taxi-taking habit and the target taxi-taking places corresponding to different taxi-taking places can be fully learned when training is carried out.
According to the embodiment, based on the historical taxi-taking information of the plurality of user samples, the selection relation between the taxi-taking habits of the user and the target taxi-taking places corresponding to different taxi-taking places can be fully excavated, and then the dependence on experience of selecting the taxi-taking places can be avoided when the selectable taxi-taking places are recommended for the user later, so that the recommendation effectiveness of the selectable taxi-taking places is improved.
In an embodiment, the extraction process of the plurality of feature sample extraction results for the current user sample may be as shown in fig. 5.
In step S501, feature extraction processing is performed on the about habit sample information of the current user sample, so as to obtain about behavior feature samples of the current user sample and about time feature samples of the current user sample.
In step S502, feature extraction processing is performed on the attribute information of the plurality of location samples of the current user sample, so as to obtain a heat feature sample and a geographic feature sample of each target boarding location corresponding to the current user sample.
In step S503, the travel characteristic sample of the current user sample, the travel time characteristic sample of the current user sample, the heat characteristic sample of each target boarding location corresponding to the current user sample, and the geographic characteristic sample are taken as a plurality of characteristic sample extraction results of the current user sample.
The specific extraction implementation process of the feature sample extraction result is similar to the specific extraction implementation process principle of the feature extraction result, and will not be described in detail herein.
In an implementation scenario, taking the deep learning model to be trained as the deep fm model as an example, based on the model structure of the deep fm model shown in fig. 6, the process of training the deep learning model to obtain the recommended prediction model may be as follows:
and loading a training set and a testing set, carrying out feature extraction by using one-hot coding to obtain a feature sample extraction result, and generating a data iterator for gradually calling an iteration deep FM model to train.
In one example, when encoding for the travel feature, 6 bits (bit) of data may be used instead, where bits 0-3 represent walking distances from the historical travel locations to the corresponding target departure locations, bit 4 represents drag behavior of the over-adjusted target departure locations, and bit 5 represents search behavior of the search target departure locations. For example: the resulting representation of the vehicle behavior feature code may be as follows: [011011] the walking distance between the historical destination and the destination is-0110, the dragging behavior of the destination is-1, and the searching behavior of the destination is-1.
When encoding the date feature of the taxi, 10 bits of data may be used instead, wherein bits 0-3 represent the date of the taxi when the user was on holiday, bits 4-7 represent the date of the taxi when the user was on weekday, and bits 8-9 represent the time interval between the current taxi and the last taxi. For example: the resulting representation of the time signature code for the about car may be as follows: [0001100020] the vehicle-closing positions of the user on holiday and on holiday are-0001, the vehicle-closing positions of the user on working day and on holiday are-1000, and the time interval between the current vehicle-closing and the last vehicle-closing is-20.
When the heat characteristic of each place attribute information corresponding to the target boarding place is coded, 9 bits of data can be used for substitution, wherein the 0 th to 2 nd bits represent the heat characteristic of the target boarding place, the 3 rd to 5 th bits represent the order taking heat characteristic of the order taking place, and the 6 th to 8 th bits represent the order taking completion heat characteristic. For example: the resulting thermal signature encoded representation may be as follows: [120110101] the heat degree characteristics of the target boarding sites are respectively-120, the ordering heat degree characteristics of the about boarding sites are-110, and the heat degree characteristics of the about boarding orders are-101.
When the geographic features of the target boarding sites corresponding to the attribute information of each site are coded, 4 bits of data can be used for substitution, wherein the 0 th bit represents the road grade of the target boarding sites, the 1 st bit represents the road width, the 2 nd bit represents the road condition of the vehicle during the restraint, and the 3 rd bit represents whether the road of the target boarding sites is forbidden or not. For example: the resulting coded representation of the geographic features may be as follows:
[1110] the road grade corresponding to the target boarding place is-1, the road width is-1, the road condition of the vehicle during vehicle restraint is-1, and whether the road where the target boarding place is located is forbidden to stop is-0.
Initializing the deep learning model and relevant super parameters, including initializing hidden layer, output layer, gradient propagation, additional gradient, loss function, training round number, training data size each time, etc. in the deep learning model. The initialization configuration includes, but is not limited to, the following settings: the hidden layer is 2 layers, the output layer is 1 layer, the training times end condition is that the error is less than 0.0001, and the training data size is 100 data samples each time.
The deep fm model may be expressed using the following formula:
Y DNN ＝σ(W |H|+1 ·a H +b |H|+1 )。
where H is the number of hidden layers.
And inputting the extraction results of the plurality of characteristic samples into the deep FM model for training until the convergence of the loss function in the deep FM model is smaller than or equal to a specified convergence threshold value, thereby obtaining an initial recommended prediction model.
Reading a verification set, inputting the verification set into the initial recommendation prediction model, verifying a training result of the initial recommendation prediction model, verifying based on an evaluation index of the model, and further repeatedly debugging the initial recommendation prediction model until the optimal recommendation prediction model is obtained and training is completed.
The evaluation index of the model may include mean square errors MSE (Mean Squared Error) and AUC (Area Under Curve), among others.
Based on the same inventive concept, the invention also provides a pushing device for the network about on-board place.
Fig. 7 is a block diagram of a pushing device for a network about a vehicle location according to an exemplary embodiment. As shown in fig. 7, the pushing device of the network about on-board place includes a first acquisition unit 701, a first determination unit 702, a recall unit 703, a screening unit 704, and a pushing unit 705.
A first obtaining unit 701, configured to obtain vehicle habit information of a user in response to receiving a vehicle constraint request triggered by the user;
A first determining unit 702, configured to determine a vehicle-restraining location when a user triggers a vehicle-restraining request;
a recall unit 703, configured to recall at least one candidate boarding location corresponding to the appointment location, and determine location attribute information of each candidate boarding location;
a screening unit 704, configured to determine an optional boarding location to be recommended from at least one candidate boarding location based on the taxi-taking habit information and location attribute information of each candidate boarding location;
and a pushing unit 705, configured to push the optional boarding location to the user, so that the user can determine the target boarding location based on the optional boarding location.
In one embodiment, the screening unit 704 includes: the first feature extraction unit is used for carrying out feature extraction processing on the vehicle habit information and the attribute information of each place respectively to obtain feature extraction results; the prediction unit is used for inputting the feature extraction result into a pre-trained recommendation prediction model and respectively determining recommendation confidence coefficient of each candidate boarding location to be recommended to a user; and the screening subunit is used for determining the optional boarding location to be recommended from at least one candidate boarding location based on the recommendation confidence degree of each candidate boarding location to be recommended to the user.
In another embodiment, the screening subunit comprises: the first screening unit is used for sequencing the candidate boarding places according to the order of the recommendation confidence coefficient of each candidate boarding place to be recommended to the user from high to low, and selecting N candidate boarding places as selectable boarding places to be recommended according to the order of the recommendation confidence coefficient from high to low, wherein N is an integer greater than 0.
In yet another embodiment, the screening subunit comprises: and the second screening unit is used for determining the candidate boarding places with the recommendation confidence degree larger than or equal to the confidence degree threshold value as the optional boarding places to be recommended according to the recommendation confidence degree of each candidate boarding place to be recommended to the user.
In yet another embodiment, the first feature extraction unit includes: the first extraction unit is used for carrying out feature extraction processing on the vehicle-restraining habit information to obtain vehicle-restraining behavior features of the user and vehicle-restraining time features of the user; the second extraction unit is used for carrying out feature extraction processing on each piece of place attribute information respectively to obtain the heat feature of each piece of place attribute information corresponding to the candidate boarding place and the geographic feature of each piece of place attribute information corresponding to the candidate boarding place; the first integration unit is used for taking the taxi-taking behavior characteristics of the user, the taxi-taking time characteristics of the user, the heat characteristics of the candidate taxi-taking places corresponding to each place attribute information and the geographic characteristics of the candidate taxi-taking places corresponding to each place attribute information as characteristic extraction results.
In one embodiment, the pushing unit 705 includes: and pushing the optional boarding places to the user according to the order of the recommendation confidence level from high to low.
In yet another embodiment, recall unit 703 comprises: the second determining unit is used for determining the area of the vehicle-restraining place; and the first recall subunit is used for recalling all boarding places in the area and serving as at least one candidate boarding place corresponding to the taxi-taking place.
In yet another embodiment, the apparatus further comprises: the third determining unit is used for determining the road where the vehicle-approaching place is located if the vehicle-approaching place does not exist in the area; and the second recall subunit is used for taking the boarding location corresponding to the road as a candidate boarding location and recalling.
The specific limitation of the pushing device for the network vehicle-on-board location can be referred to the limitation of the pushing method for the network vehicle-on-board location, and will not be described herein. The various modules described above may be implemented in whole or in part by software, hardware, or a combination thereof. The above modules may be embedded in hardware or independent of a processor in the electronic device, or may be stored in software in a memory in the electronic device, so that the processor may call and execute operations corresponding to the above modules.
Based on the same inventive concept, the invention also provides a training device of the prediction model.
Fig. 8 is a block diagram of a training apparatus for a predictive model according to an exemplary embodiment. As shown in fig. 8, the training apparatus of the prediction model includes a second acquisition unit 801, a second feature extraction unit 802, and a training unit 803.
The second obtaining unit 801 is configured to obtain historical taxi-taking information of a plurality of user samples within a specified time range, where the historical taxi-taking information includes taxi-taking habit sample information corresponding to the user samples and location sample attribute information of a target taxi-taking location selected by each taxi-taking.
The second feature extraction unit 802 is configured to perform feature extraction processing on the habit sample information of each user sample and the corresponding location sample attribute information, so as to obtain a plurality of feature sample extraction results.
The training unit 803 is configured to train the deep learning model based on the plurality of feature sample extraction results, and obtain a recommendation prediction model, where the recommendation prediction model is used to predict a recommendation confidence level of the candidate boarding location to be recommended to the user.
In an embodiment, the second feature extraction unit 802 includes: the third extraction unit is used for carrying out feature extraction processing on the taxi-taking habit sample information of the current user sample to obtain a taxi-taking behavior feature sample of the current user sample and a taxi-taking time feature sample of the current user sample; the fourth extraction unit is used for carrying out feature extraction processing on the attribute information of the plurality of place samples of the current user sample respectively to obtain a heat feature sample and a geographic feature sample of each target boarding place corresponding to the current user sample; the second integration unit is used for taking the taxi-taking behavior feature sample of the current user sample, the taxi-taking time feature sample of the current user sample, the heat feature sample of each target taxi-taking place corresponding to the current user sample and the geographic feature sample as a plurality of feature sample extraction results of the current user sample.
The specific limitation of the training device of the prediction model and the beneficial effects can be referred to the limitation of the training method of the prediction model, and are not repeated herein. The various modules described above may be implemented in whole or in part by software, hardware, or a combination thereof. The above modules may be embedded in hardware or independent of a processor in the electronic device, or may be stored in software in a memory in the electronic device, so that the processor may call and execute operations corresponding to the above modules.
Fig. 9 is a schematic diagram of a hardware structure of an electronic device according to an exemplary embodiment. As shown in fig. 9, the device includes one or more processors 910 and a memory 920, where the memory 920 includes persistent memory, volatile memory, and a hard disk, one processor 910 being illustrated in fig. 9. The apparatus may further include: an input device 930, and an output device 940.
The processor 910, memory 920, input device 930, and output device 940 may be connected by a bus or other means, for example in fig. 9.
The processor 910 may be a central processing unit (Central Processing Unit, CPU). The processor 910 may also be any other general purpose processor, digital signal processor (Digital Signal Processor, DSP), application specific integrated circuit (Application Specific Integrated Circuit, ASIC), field programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory 920, which is a non-transitory computer readable storage medium, includes persistent memory, volatile memory, and a hard disk, may be used to store non-transitory software programs, non-transitory computer executable programs, and modules, such as program instructions/modules corresponding to the service management method in the embodiments of the present application. The processor 910 executes various functional applications and data processing of the server by running non-transitory software programs, instructions and modules stored in the memory 920, that is, implementing any of the above-mentioned pushing methods of the network-bound vehicle on-vehicle location or training methods of the prediction model.
The input device 930 may receive input numeric or character information and generate key signal inputs related to user settings and function control. The output device 940 may include a display device such as a display screen.
One or more modules are stored in the memory 920 that, when executed by the one or more processors 910, perform the methods illustrated in fig. 1-6.
The product can execute the method provided by the embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method. Technical details which are not described in detail in the present embodiment can be found in the embodiments shown in fig. 1 to 6.
The embodiment of the invention also provides a non-transitory computer storage medium, wherein the computer storage medium stores computer executable instructions, and the computer executable instructions can execute the authentication method in any of the method embodiments. The storage medium may be a magnetic Disk, an optical Disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a Flash Memory (Flash Memory), a Hard Disk (HDD), or a Solid State Drive (SSD); the storage medium may also comprise a combination of memories of the kind described above.
It is apparent that the above examples are given by way of illustration only and are not limiting of the embodiments. Other variations or modifications of the above teachings will be apparent to those of ordinary skill in the art. It is not necessary here nor is it exhaustive of all embodiments. While still being apparent from variations or modifications that may be made by those skilled in the art are within the scope of the invention.
Claims (14)
1. A push method for a network about on-board location, the method comprising:
responding to a vehicle-restraining request triggered by a user, and acquiring vehicle-restraining habit information of the user;
determining a vehicle-restraining place when the user triggers the vehicle-restraining request;
recalling at least one candidate boarding location corresponding to the taxi-taking location, and determining location attribute information of each candidate boarding location;
determining an optional boarding location to be recommended from the at least one candidate boarding location based on the taxi-habit information and location attribute information of each candidate boarding location;
pushing the optional boarding location to the user for the user to determine a target boarding location based on the optional boarding location.
2. The method of claim 1, wherein the determining an alternative boarding location to be recommended from the at least one candidate boarding location based on the taxi-habit information and location attribute information of each candidate boarding location comprises:
respectively carrying out feature extraction processing on the vehicle habit information and the attribute information of each place to obtain a feature extraction result;
inputting the feature extraction result into a pre-trained recommendation prediction model, and respectively determining recommendation confidence of each candidate boarding location to be recommended to the user;
and determining optional boarding places to be recommended from the at least one candidate boarding place based on recommendation confidence of each candidate boarding place to be recommended to the user.
3. The method of claim 2, wherein the determining the alternative pick-up location to be recommended from the at least one candidate pick-up location based on the confidence of the recommendation to be recommended to the user for each candidate pick-up location comprises:
and sequencing the candidate boarding places according to the order of the recommendation confidence coefficient of each candidate boarding place to be recommended to the user from high to low, and selecting N candidate boarding places as selectable boarding places to be recommended according to the order of the recommendation confidence coefficient from high to low, wherein N is an integer greater than 0.
4. The method of claim 2, wherein the determining the alternative pick-up location to be recommended from the at least one candidate pick-up location based on the confidence of the recommendation to be recommended to the user for each candidate pick-up location comprises:
and determining the candidate boarding places with the recommendation confidence coefficient larger than or equal to the confidence coefficient threshold value as the selectable boarding places to be recommended according to the recommendation confidence coefficient of each candidate boarding place to be recommended to the user.
5. The method according to any one of claims 2 to 4, wherein the performing feature extraction processing on the vehicle habit information and each location attribute information to obtain feature extraction results includes:
performing feature extraction processing on the vehicle-restraining habit information to obtain vehicle-restraining behavior features of the user and vehicle-restraining time features of the user;
respectively carrying out feature extraction processing on each place attribute information to obtain the heat feature of each place attribute information corresponding to the candidate boarding place and the geographic feature of each place attribute information corresponding to the candidate boarding place;
and taking the feature of the taxi-taking behavior of the user, the feature of the taxi-taking time of the user, the heat feature of each place attribute information corresponding to the candidate taxi-taking place and the geographic feature of each place attribute information corresponding to the candidate taxi-taking place as feature extraction results.
6. The method of claim 3 or 4, wherein the pushing the optional boarding location to the user comprises:
and pushing the optional boarding places to the user according to the sequence of the recommendation confidence from high to low.
7. The method of claim 1, wherein the recalling the at least one candidate pick-up location corresponding to the about location comprises:
determining the area of the vehicle-restraining place;
and recalling all boarding places in the area as at least one candidate boarding place corresponding to the taxi-taking place.
8. The method of claim 7, wherein the method further comprises:
if no boarding location exists in the area, determining a road where the taxi-taking location is located;
and taking the boarding location corresponding to the road as a candidate boarding location and recalling.
9. A method of training a predictive model, the method comprising:
acquiring historical taxi-taking information of a plurality of user samples within a specified time range, wherein the historical taxi-taking information comprises taxi-taking habit sample information corresponding to the user samples and place sample attribute information of a target taxi-taking place selected by each taxi-taking;
Respectively carrying out feature extraction processing on the vehicle habit sample information of each user sample and the corresponding place sample attribute information to obtain a plurality of feature sample extraction results;
training a deep learning model based on the extraction results of the plurality of feature samples to obtain a recommendation prediction model, wherein the recommendation prediction model is used for predicting recommendation confidence of a candidate on-board location to be recommended to a user.
10. The method of claim 9, wherein the performing feature extraction processing on the about habit sample information and the corresponding location sample attribute information of each user sample to obtain a plurality of feature sample extraction results includes:
feature extraction processing is carried out on the taxi-taking habit sample information of the current user sample, and a taxi-taking behavior feature sample of the current user sample and a taxi-taking time feature sample of the current user sample are obtained;
respectively carrying out feature extraction processing on the attribute information of the plurality of place samples of the current user sample to obtain a heat feature sample and a geographic feature sample of each target boarding place corresponding to the current user sample;
and taking the taxi-taking behavior feature sample of the current user sample, the taxi-taking time feature sample of the current user sample, the heat feature sample of each target boarding place corresponding to the current user sample and the geographic feature sample as a plurality of feature sample extraction results of the current user sample.
11. A push device for a network restraint on-board location, the device comprising:
the first acquisition unit is used for responding to a vehicle-restraining request triggered by a user and acquiring vehicle-restraining habit information of the user;
the first determining unit is used for determining a vehicle-restraining place when the user triggers the vehicle-restraining request;
a recall unit, configured to recall at least one candidate boarding location corresponding to the taxi-taking location, and determine location attribute information of each candidate boarding location;
a screening unit, configured to determine an optional boarding location to be recommended from the at least one candidate boarding location based on the taxi-taking habit information and location attribute information of each candidate boarding location;
and the pushing unit is used for pushing the optional boarding location to the user so that the user can determine a target boarding location based on the optional boarding location.
12. A training device for a predictive model, the training device comprising:
the second acquisition unit is used for acquiring historical taxi-taking information of a plurality of user samples within a specified time range, wherein the historical taxi-taking information comprises taxi-taking habit sample information corresponding to the user samples and place sample attribute information of a target taxi-taking place selected by each taxi-taking;
The second feature extraction unit is used for carrying out feature extraction processing on the about habit sample information of each user sample and the corresponding place sample attribute information respectively to obtain a plurality of feature sample extraction results;
the training unit is used for training the deep learning model based on the extraction results of the plurality of characteristic samples to obtain a recommendation prediction model, and the recommendation prediction model is used for predicting recommendation confidence of the candidate boarding location to be recommended to the user.
13. An electronic device comprising a memory and a processor, the memory and the processor being communicatively connected to each other, the memory having stored therein computer instructions, the processor executing the computer instructions to perform the method of pushing a network vehicle location according to any one of claims 1-8 or the method of training a predictive model according to any one of claims 9-10.
14. A computer readable storage medium storing computer instructions for causing the computer to perform the push method of the network about a vehicle location as claimed in any one of claims 1 to 8 or the training method of the predictive model as claimed in any one of claims 9 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310150283.7A CN116340600A (en) | 2023-02-13 | 2023-02-13 | Push method of network appointment vehicle on-vehicle location, training method and device of prediction model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310150283.7A CN116340600A (en) | 2023-02-13 | 2023-02-13 | Push method of network appointment vehicle on-vehicle location, training method and device of prediction model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116340600A true CN116340600A (en) | 2023-06-27 |
Family
ID=86890654
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310150283.7A Pending CN116340600A (en) | 2023-02-13 | 2023-02-13 | Push method of network appointment vehicle on-vehicle location, training method and device of prediction model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116340600A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111861647A (en) * | 2020-07-03 | 2020-10-30 | 北京嘀嘀无限科技发展有限公司 | Method and system for recommending boarding points |
| CN112270427A (en) * | 2020-11-10 | 2021-01-26 | 北京嘀嘀无限科技发展有限公司 | Method and system for recommending boarding points |
| CN113159396A (en) * | 2021-03-31 | 2021-07-23 | 广州宸祺出行科技有限公司 | Adaptive adsorption method and system for recommending boarding points |
| CN114037589A (en) * | 2016-06-13 | 2022-02-11 | 北京嘀嘀无限科技发展有限公司 | Boarding point recommendation processing method and system |
-
2023
- 2023-02-13 CN CN202310150283.7A patent/CN116340600A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114037589A (en) * | 2016-06-13 | 2022-02-11 | 北京嘀嘀无限科技发展有限公司 | Boarding point recommendation processing method and system |
| CN111861647A (en) * | 2020-07-03 | 2020-10-30 | 北京嘀嘀无限科技发展有限公司 | Method and system for recommending boarding points |
| CN112270427A (en) * | 2020-11-10 | 2021-01-26 | 北京嘀嘀无限科技发展有限公司 | Method and system for recommending boarding points |
| CN113159396A (en) * | 2021-03-31 | 2021-07-23 | 广州宸祺出行科技有限公司 | Adaptive adsorption method and system for recommending boarding points |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107945507B (en) | Travel time prediction method and device | |
| US9958288B2 (en) | Planning under destination uncertainty | |
| US10852150B2 (en) | Methods and apparatuses for fuel consumption prediction | |
| WO2016111857A1 (en) | Predicting and utilizing variability of travel times in mapping services | |
| WO2021121354A1 (en) | Model-based deep reinforcement learning for dynamic pricing in online ride-hailing platform | |
| CN109360421B (en) | Traffic information prediction method and device based on machine learning and electronic terminal | |
| US20210199449A1 (en) | System and method for alternative destination recommendation on ridesharing platforms | |
| US20240027206A1 (en) | Transportation route error detection and adjustment | |
| CN114218483A (en) | Parking recommendation method and application thereof | |
| CN111859172B (en) | Information pushing method, device, electronic equipment and computer readable storage medium | |
| US20220327650A1 (en) | Transportation bubbling at a ride-hailing platform and machine learning | |
| CN116340600A (en) | Push method of network appointment vehicle on-vehicle location, training method and device of prediction model | |
| CN109034494B (en) | Bus dispatching method | |
| CN111507541A (en) | Goods quantity prediction model construction method, goods quantity measurement device and electronic equipment | |
| WO2020254418A1 (en) | System and method for populating a database with occupancy data of parking facilities | |
| CN115953226A (en) | Vehicle rental risk assessment method and system | |
| CN110910191A (en) | Car pooling order generation method and equipment | |
| CN113570004B (en) | Riding hot spot area prediction method, device, equipment and readable storage medium | |
| CN110211381B (en) | Traffic route distribution method and device | |
| CN114399351A (en) | Information recommendation method and device, electronic equipment and storage medium | |
| US10955251B2 (en) | Identifying incorrect coordinate prediction using route information | |
| CN112304326A (en) | Travel route planning method, equipment and storage medium | |
| CN111813881A (en) | Method, apparatus, device and storage medium for travel information processing | |
| CN116798234B (en) | Method, device, computer equipment and storage medium for determining station parameter information | |
| CN117330094A (en) | Route recommendation method, device, equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |