CN111861538A

CN111861538A - Information pushing method and device, electronic equipment and storage medium

Info

Publication number: CN111861538A
Application number: CN202010491309.0A
Authority: CN
Inventors: 万怀宇; 单明辉; 曹仲; 郭晟楠; 徐吉兴; 脱立恒; 林友芳
Original assignee: Beijing Jiaotong University; Beijing Didi Infinity Technology and Development Co Ltd
Current assignee: Beijing Jiaotong University; Beijing Didi Infinity Technology and Development Co Ltd
Priority date: 2020-06-02
Filing date: 2020-06-02
Publication date: 2020-10-30
Anticipated expiration: 2040-06-02
Also published as: CN111861538B

Abstract

The application provides an information pushing method, an information pushing device, electronic equipment and a storage medium, wherein the method comprises the following steps: inputting historical trip information of a target user into a trip prediction model to determine first planned trip information of the target user in a target time period; selecting a target trip resource from a plurality of candidate trip resources according to the first planned trip information and resource attribute information of the candidate trip resources; and pushing the target trip resources to the target user so as to improve the utilization rate of the pushed information.

Description

Information pushing method and device, electronic equipment and storage medium

Technical Field

The present application relates to the field of information technologies, and in particular, to an information pushing method and apparatus, an electronic device, and a storage medium.

Background

In recent years, with the rapid development of travel platforms, various travel platforms are dedicated to promote traffic resource sharing and improve the service quality of the travel platforms.

In order to improve the service quality of the travel platform, the travel platform can push travel resource information to platform users, and when the travel resource information is pushed, indiscriminate pushing can be carried out on the users of the platform, but the users can continuously receive the travel resource information which is not concerned by the users, so that the experience degree of the users is reduced, the travel resource waste can be caused, and the utilization rate of the pushed travel resource information is reduced.

Disclosure of Invention

In view of the above, an object of the present application is to provide an information pushing method, an information pushing apparatus, an electronic device, and a storage medium, so as to improve utilization rate of pushed information.

In a first aspect, an embodiment of the present application provides an information pushing apparatus, where the apparatus includes:

the determining module is used for inputting the historical trip information of the target user into the trip prediction model so as to determine first planned trip information of the target user in a target time period;

the selection module is used for selecting a target trip resource from a plurality of candidate trip resources according to the first planned trip information and the resource attribute information of the candidate trip resources;

and the pushing module is used for pushing the target trip resources to the target user.

In one embodiment, the selection module is configured to select the target travel resource according to the following steps:

acquiring historical travel information of each reference user in a reference user group;

for each reference user, inputting historical travel information of the reference user into the travel prediction model to determine second planned travel information of the reference user in a target time period;

and selecting a target trip resource from a plurality of candidate trip resources according to the second planned trip information, the first planned trip information and the resource attribute information of the candidate trip resources.

In one embodiment, the selecting module is configured to select the target travel resource according to the following steps:

if the resource attribute information of the trip resource exists in the plurality of candidate trip resources and is matched with the trip attribute information of the first planned trip information, determining the trip resource of which the resource attribute information is matched with the trip attribute information of the first planned trip information as the target trip resource.

In one embodiment, the determining module is configured to determine the first planned travel information of the target user at the target time period according to the following steps:

determining a candidate starting and ending point group set corresponding to the target user based on the historical travel order of the target user;

for each candidate start-stop point group in the candidate start-stop point group set, inputting both historical travel information of a target user and predicted travel information corresponding to the candidate start-stop point group into the travel prediction model, and obtaining the selection probability of the target user for selecting the candidate start-stop point group in a target time period;

selecting a target starting and end point pair from the candidate starting and end point group set based on the selection probability corresponding to each candidate starting and end point group;

and determining first planned travel information of the target user in a target time period according to the selected target starting and ending point pair.

In one embodiment, the determining module is configured to determine the probability that the target user selects the candidate starting and ending point in the target time period according to the following steps:

determining a trip characteristic vector corresponding to the candidate starting and ending point group according to the predicted trip information corresponding to the candidate starting and ending point group;

generating a historical travel characteristic vector corresponding to each historical travel order of the target user based on historical travel information corresponding to the historical travel order;

inputting a historical trip characteristic vector corresponding to the historical trip order and a trip behavior characteristic vector corresponding to a previous historical trip order adjacent to the historical trip order into a characteristic vector generation layer in the trip prediction model to obtain a trip behavior characteristic vector corresponding to the historical trip order;

inputting the travel behavior characteristic vector corresponding to the historical travel order and the travel characteristic vector corresponding to the candidate starting and ending point group into an attention layer in the travel prediction model to obtain an influence coefficient of the historical travel order on the candidate starting and ending point group;

determining a first feature vector corresponding to the target user based on an influence coefficient of each historical travel order on the candidate starting and ending point group and a travel behavior feature vector corresponding to each historical travel order;

And inputting the first feature vector corresponding to the target user and the trip feature vector corresponding to the candidate starting and ending point group into a probability prediction layer in the trip prediction model to obtain the probability that the target user selects the candidate starting and ending point in the target time period.

In one embodiment, the target time period is determined according to the following steps:

acquiring historical travel orders of a reference user group;

determining a travel frequency corresponding to each time period in a plurality of preset time periods based on travel time in a historical travel order of a reference user group;

and determining the target time period from a plurality of time periods according to the travel frequency corresponding to each time period.

In one embodiment, the determining module is configured to determine the planned travel information of the target user in the target time period according to the following steps, including:

acquiring the road congestion degree of a target time period;

and if the road congestion degree of the target time period exceeds a preset value, inputting historical travel information of the target user into a travel prediction model so as to determine first planned travel information of the target user in the target time period.

In one embodiment, the historical travel information includes any one of the following information:

Historical trip starting and ending point information, historical trip time information, historical trip weather information, historical trip starting and ending point region distribution information and historical trip time difference information.

In one embodiment, the target travel resource includes any one of the following resources:

preferential resources are taken out; and (5) travel mode resources.

In a second aspect, an embodiment of the present application provides an information pushing method, where the method includes:

inputting historical trip information of a target user into a trip prediction model to determine first planned trip information of the target user in a target time period;

selecting a target trip resource from a plurality of candidate trip resources according to the first planned trip information and resource attribute information of the candidate trip resources;

and pushing the target trip resources to the target user.

In one embodiment, selecting a target trip resource from a plurality of candidate trip resources comprises:

In one embodiment, selecting a target trip resource from a plurality of candidate trip resources according to the first planned trip information and resource attribute information of the candidate trip resource includes:

In one embodiment, determining the first planned travel information of the target user in the target time period includes:

In one embodiment, for each candidate starting and ending point group in the candidate starting and ending point group set, inputting historical travel information of a target user and predicted travel information corresponding to the candidate starting and ending point group into the travel prediction model, and determining a probability that the target user selects the candidate starting and ending point in a target time period includes:

acquiring historical travel orders of a reference user group;

In one embodiment, inputting the historical travel information of the target user into a travel prediction model to determine first planned travel information of the target user in a target time period includes:

Acquiring the road congestion degree of a target time period;

historical trip starting and ending point information, historical trip time information, historical trip weather information, historical trip starting and ending point area information and historical trip time difference information.

preferential resources are taken out; and (5) travel mode resources.

In a third aspect, an embodiment of the present application provides an electronic device, including: the information pushing method comprises a processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, when an electronic device runs, the processor and the storage medium communicate through the bus, and the processor executes the machine-readable instructions to execute the steps of the information pushing method.

In a fourth aspect, the present application provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the computer program performs the steps of the information pushing method.

According to the information pushing method provided by the embodiment of the application, historical trip information of a target user is considered, the historical trip information of the target user is input into a trip prediction model, first planned trip information of the target user in a target time period is obtained, the accuracy of the obtained planned trip information is improved to a certain extent, a target trip resource is selected for the target user from a plurality of candidate trip resources according to the first planned trip information and resource attribute information of preset candidate trip resources, the matching degree between the target trip resource and the target user is improved, the target trip resource is pushed to the target user, and the utilization rate of the pushed trip resource is improved.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained from the drawings without inventive effort.

Fig. 1 shows an architecture diagram of an information push system provided by an embodiment of the present application;

fig. 2 shows a first flowchart of an information pushing method provided by an embodiment of the present application;

fig. 3 shows a second flowchart of an information pushing method provided by an embodiment of the present application;

fig. 4 is a schematic structural diagram illustrating an information pushing apparatus provided by an embodiment of the present application;

fig. 5 shows a schematic structural diagram of an electronic device provided in an embodiment of the present application.

Detailed Description

In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it should be understood that the drawings in the present application are for illustrative and descriptive purposes only and are not used to limit the scope of protection of the present application. Additionally, it should be understood that the schematic drawings are not necessarily drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of the present application. It should be understood that the operations of the flow diagrams may be performed out of order, and steps without logical context may be performed in reverse order or simultaneously. One skilled in the art, under the guidance of this application, may add one or more other operations to, or remove one or more operations from, the flowchart.

In addition, the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. The components of the embodiments of the present application, generally described and illustrated in the figures herein, can be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present application, presented in the accompanying drawings, is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present application without making any creative effort, shall fall within the protection scope of the present application.

To enable those skilled in the art to use the present disclosure, the following embodiments are presented in conjunction with a specific application scenario, "travel scenario". It will be apparent to those skilled in the art that the general principles defined herein may be applied to other embodiments and applications without departing from the spirit and scope of the application. Although the present application primarily focuses on travel scenarios, it should be understood that this is only one exemplary embodiment.

It should be noted that in the embodiments of the present application, the term "comprising" is used to indicate the presence of the features stated hereinafter, but does not exclude the addition of further features.

The terms "passenger," "requestor," "service requestor," and "customer" are used interchangeably in this application to refer to an individual, entity, or tool that can request or order a service. The terms "driver," "provider," "service provider," and "provider" are used interchangeably in this application to refer to an individual, entity, or tool that can provide a service. The term "user" in this application may refer to an individual, entity or tool that requests a service, subscribes to a service, provides a service, or facilitates the provision of a service. For example, the user may be a passenger, a driver, an operator, etc., or any combination thereof. In the present application, "passenger" and "passenger terminal" may be used interchangeably, and "driver" and "driver terminal" may be used interchangeably.

The terms "service request" and "order" are used interchangeably herein to refer to a request initiated by a passenger, a service requester, a driver, a service provider, or a supplier, the like, or any combination thereof. Accepting the "service request" or "order" may be a passenger, a service requester, a driver, a service provider, a supplier, or the like, or any combination thereof. The service request may be charged or free.

In order to improve the utilization rate of the pushed travel resource, the travel information of the passenger can be predicted based on macroscopic traffic data, such as regional traffic flow and road flow demand, however, the influence factors influencing the travel of the passenger are more, such as weather factors, personal preference factors of the passenger, working factors, traffic environment factors and the like, the travel of the passenger is predicted only by considering the traffic data, the accuracy of the predicted travel is lower, when the travel resource information is pushed to the passenger based on the predicted travel, the pushed resource information is not necessarily the travel resource information required by the passenger, and the utilization rate of the travel resource is not improved by the travel resource information pushed in the above manner.

Based on the above, the application provides an information pushing method, historical trip information of a target user is considered, the historical trip information of the target user is input into a trip prediction model, first planned trip information of the target user in a target time period is obtained, accuracy of the obtained planned trip information is improved to a certain extent, a target trip resource is selected for the target user from a plurality of candidate trip resources according to the first planned trip information and resource attribute information of preset candidate trip resources, matching degree between the target trip resource and the target user is improved, the target trip resource is pushed to the target user, utilization rate of the pushed trip resource is improved, and service quality of a trip platform is also improved.

Fig. 1 is a schematic structural diagram of an information push system according to an embodiment of the present application. For example, the information push system may be an online transportation service platform for transportation services such as taxi cab, designated drive service, express, carpool, bus service, driver rental, or regular service, or any combination thereof. The information push system may include one or more of a server 110, a network 120, a service requester terminal 130, a service provider terminal 140, and a database 150.

In some embodiments, the server 110 may include a processor. The processor may process information and/or data related to the service request to perform one or more of the functions described herein. For example, the processor may determine the target vehicle based on a service request obtained from the service requester terminal 130. In some embodiments, a processor may include one or more processing cores (e.g., a single-core processor (S) or a multi-core processor (M)). Merely by way of example, a Processor may include a Central Processing Unit (CPU), an Application Specific Integrated Circuit (ASIC), an Application specific Instruction set Processor (ASIP), a Graphics Processing Unit (GPU), a Physical Processing Unit (PPU), a Digital Signal Processor (DSP), a Field Programmable Gate Array (FPGA), a Programmable Logic Device (PLD), a controller, a microcontroller Unit, a Reduced Instruction set computer (Reduced Instruction set computing), a microprocessor, or the like, or any combination thereof.

In some embodiments, the device types corresponding to the service requester terminal 130 and the service provider terminal 140 may be mobile devices, such as smart home devices, wearable devices, smart mobile devices, virtual reality devices, augmented reality devices, and the like, and may also be tablet computers, laptop computers, built-in devices in motor vehicles, and the like.

In some embodiments, a database 150 may be connected to the network 120 to communicate with one or more components in the information push system (e.g., the server 110, the service requester terminal 130, the service provider terminal 140, etc.). One or more components in the information push system may access data or instructions stored in the database 150 via the network 120. In some embodiments, the database 150 may be directly connected to one or more components in the information push system, or the database 150 may be part of the server 110.

The information push method provided by the embodiment of the present application is described in detail below with reference to the content described in the information push system shown in fig. 1.

An embodiment of the present application provides an information pushing method, as shown in fig. 2, where the method is applied to a server of a travel service platform, and the method specifically includes the following steps:

S201, inputting historical trip information of the target user into a trip prediction model to determine first planned trip information of the target user in a target time period.

S202, selecting a target trip resource from the plurality of candidate trip resources according to the first planned trip information and the resource attribute information of the candidate trip resources.

S203, pushing the target trip resource to the target user.

In S201, the historical travel information is determined based on a plurality of historical travel orders of the target user.

The historical trip information comprises at least one of historical trip starting and ending point information, historical trip time information, historical trip weather information, historical trip starting and ending point region distribution information, historical trip time difference information and the like.

The historical trip start and end point information includes historical trip start point information and historical trip end point information, for example, the historical trip start point information may be a trip start point identifier in the historical trip order, and the historical trip end point information may be a trip end point identifier in the historical trip order. The start and end points in this application indicate the departure and destination of a user's trip, e.g., r_u,i＝<t_u,i,o_u,i,d_u,iRepresents the ith trip of user u, time is t _u,iThe origin (starting point) is o_u,iThe destination (end point) is d_u,iAnd M user sets are denoted as U ═ U₁,u₂,…,u_M}，u_MIndicating the mth user.

The historical travel time information may be an order-issuing time in the historical travel order, or a travel time of the user.

The historical trip weather information is weather conditions of the user when the user trips, such as whether the user falls into rain, temperature information, humidity information and the like.

The historical trip start-end point region distribution information comprises historical trip start-point region distribution information and historical trip end point region distribution information, the historical trip start-point region distribution information is distribution information of each interest point of a region where a trip start point is located in a historical trip order, namely, the distribution situation of the interest points in the region where the trip start point is located, for example, the region where the historical trip start point of a user is located is a central village, the interest points can be catering places, houses, office buildings and the like, and the distribution situation is the distribution situation of the catering places, houses, office buildings and the like in the central village region; the historical trip end point region distribution information is distribution information of each interest point of a region where a trip end point is located in a historical trip order, that is, distribution information of the interest points in the region where the trip end point is located, for example, the region where the historical trip end point of the user is located is an upper region, the interest points can be schools, houses, office buildings and the like, and the distribution conditions of the schools, the houses and the office buildings are the upper region; the starting point region and the ending point region may be obtained by mesh division or road network division, and for example, the N region sets are expressed as P ═ { P ═ ₁,p₂,…,p_NDenotes that the urban area is divided into N areas, p_NRepresents the Nth area, and thus, there is N for one trip of the user²And (4) carrying out the following steps.

The historical travel time difference information represents a time difference between the travel time of the current historical travel order and the travel time of the previous historical travel order, or an absolute value of the time difference between the travel time of the current historical travel order and the travel time of the next historical travel order, and can be determined according to actual conditions. When the historical travel time difference information is determined, the historical travel orders of the user can be sequenced according to the sequence of the travel time from far to near, the previous historical travel order is the previous historical travel order of the current historical travel order in the sequence, and the next historical travel order is the next historical travel order of the current historical travel order in the sequence.

The target time period is a time period after the time point of the planned travel information is determined, the time period may be a time interval in any date after the time point of the planned travel information is determined, and may also be consecutive days, and the target time period may be an arbitrarily selected time period, for example, the time point of the planned travel information of the target user is determined to be 20 days in 12 months in 2019, and 21 days in 12 months in 2019-7: 00-9: 00.

In a time period corresponding to any selection in a historical date, travel behaviors of a user are less, so that travel characteristics of the user are sparse, in order to obtain a more targeted target time period, historical travel time of the user is considered, so that a time period with a higher travel frequency is selected, and the time period with the higher travel frequency is used as the target time period, and the method specifically comprises the following steps:

acquiring historical travel orders of a reference user group;

Here, the reference user group includes a target user and a large number of other users; the historical travel orders are finished historical travel orders of the reference users in the reference user group, and the travel time is the order issuing time of the users included in the historical travel orders or the departure time of the users; the trip frequency may be the number of historical trip orders corresponding to the trip time falling into the time period, and the higher the number is, the more historical trip orders represented in the time period are.

In a specific implementation process, historical travel orders corresponding to a large number of users are obtained, for each historical travel order, a time period including the travel time is determined according to the travel time in the historical travel order, the number of the historical travel orders of which the travel time falls into each time period is counted, the number is used as the travel frequency of the corresponding time period, the time period corresponding to the maximum travel frequency can be determined as a target time period, the travel orders can be sequenced from large to small according to the travel frequency, and the time periods corresponding to the front preset number of travel frequencies sequenced in the front are determined as the target time period. The preset number can be determined according to actual conditions.

When determining the first planned travel information of the target user in the target time period, the first planned travel information of the target user in the target time period may be determined at a preset time point, and whether to trigger the determination of the first planned travel information of the target user in the target time period may also be determined according to a road congestion degree, specifically including the following steps:

the method comprises the steps of obtaining road congestion degree of a target time period, and inputting historical travel information of a target user into a travel prediction model if the road congestion degree of the target time period exceeds a preset value so as to determine first planned travel information of the target user in the target time period.

The road congestion degree can be represented by a road congestion index, and the higher the road congestion index is, the more serious the road congestion degree is; the preset value is determined according to the degree of congestion of the historical road.

In the specific implementation process, after the road congestion degree of each road in the road network in the target time period is obtained, the road congestion index and the preset numerical value of each road in the road network are compared, if the road congestion index of each road is larger than the preset numerical value, the road identification of each road is recorded, the number of the road identifications with the road congestion indexes larger than the preset numerical value is counted, the ratio of the number to the total number of the roads in the road network is calculated to be larger than the preset proportional threshold, the step of inputting the historical trip information of the target user into the trip prediction model to determine the first planned trip information of the target user in the target time period is executed, and therefore, compared with the step of calculating at a specific time point, the calculation times of the server are reduced to a certain extent.

The trip prediction model includes a feature vector generation Layer, an Attention Layer, a probability prediction Layer, and the like, where the feature vector generation Layer, the Attention Layer, and the probability prediction Layer are connected in sequence, the feature vector prediction Layer may include a Gated Recurrent Unit (GRU) network, the Attention Layer may be a neural network Attention mechanism (Attention) Layer, and the probability prediction Layer may include a Multi-Layer Perceptron (MLP). The target time periods of planned travel information obtained by prediction of different travel prediction models are different, and for one target prediction model, the model can predict the planned travel information of a user in one target time period.

The travel prediction model is obtained by training based on first historical travel information of a large number of users, the first historical travel information is determined from historical travel orders of the users, and the historical travel orders can be the same as or different from the historical travel orders of the target users and are determined according to actual conditions.

The training process of the travel prediction model is detailed below.

The method comprises the steps of obtaining a historical travel order of each user in a reference user group, and determining first historical travel information of each user, actual travel information of the user in a historical target time period, first historical interest area information corresponding to the user, and a historical candidate starting and ending point group set corresponding to the user based on the historical travel order of each user.

The actual travel information of the user in the historical target time period is an actual starting point and an actual end point of the user in the historical target time period, and the determination process of the historical target time period may refer to the determination process of the target time period.

The first historical travel information includes at least one of first historical travel starting and ending point information, first historical travel time information, first historical travel weather information, first historical travel starting and ending point region distribution information, first historical travel time difference information and the like, and the meaning of the information in the first historical travel information can refer to the meaning of the information in the historical travel information.

The first historical interest area information may include a first historical interest starting point area identifier, a first historical interest end point area identifier, a first historical interest starting point pair identifier, and the like, where the first historical interest starting point pair identifier may be determined based on starting point pairs in the historical travel order, for example, counting the number of the same starting point pairs (the starting points are the same, and the end points are the same) in the historical travel order, sorting the starting point pairs according to the order of the number from large to small, and using the top-ranked preset number of starting point pairs as the interest starting point pair identifiers; the first historical interest starting point area identification is determined based on a starting point and an end point in the historical travel orders of the user, for each historical travel order, a starting point area where the starting point is located and an end point area where the end point is located in the historical travel order are respectively determined, a first number of the same starting point areas and a second number of the same end point areas are counted, the starting point area pairs are sorted according to the sequence of the first number from large to small, the starting point areas with the preset number which are sorted in the front are used as interest starting point areas, the end point area pairs are sorted according to the sequence of the second number from large to small, and the end point areas with the preset number which are sorted in the front are used as interest starting point areas.

The historical starting and ending point candidate group set may be selected from starting and ending point pairs of the historical travel order, for example, the number of the same starting and ending point pairs is determined, the starting and ending point pairs are sorted in descending order of number, and a preset number of starting and ending point pairs with the top sorting order are determined as the historical starting and ending point candidate group set.

And constructing a training set, wherein the training set comprises first historical travel information corresponding to each historical travel order corresponding to each user, second historical travel information corresponding to a historical candidate starting and ending point group in a new first historical interest area and historical candidate starting and ending point group set, and actual travel information of each user in a historical target time period. The second historical trip information includes at least one of second historical trip starting and ending point information, second historical trip time information, second historical trip weather information, second historical trip starting and ending point region distribution information, second historical trip time difference information and the like, and the meaning of the information in the second historical trip information can refer to the meaning of the information in the historical trip information.

According to the first historical travel information corresponding to each historical travel order, determining a user characteristic, an order characteristic, a weather characteristic, a starting point characteristic, an end point characteristic, a time characteristic, a starting point area distribution characteristic and an end point area distribution characteristic corresponding to each historical travel order, and performing characteristic combination under different characteristics on the user characteristic, the order characteristic, the weather characteristic, the starting point and end point characteristic, the time characteristic, the starting point area distribution characteristic and the end point area distribution characteristic to obtain a combination characteristic corresponding to the historical travel order. For example, the starting point feature and the starting point region distribution feature are subjected to feature combination to obtain a first feature, the ending point feature and the ending point region distribution feature are subjected to feature combination to obtain a second feature, and a value under the first feature and a value under the second feature are further subjected to weighted calculation.

The characteristics are expressed as follows: the ith group of users is characterized as

Identifying ith in group characteristics_iVector representation of individual features,/_iRepresenting the number of features in the vector i, K_iThe sum of the dimensions of all features representing the ith group feature. It should be noted that the interval feature in the order feature is expressed as

Wherein t is_iThe departure time of the ith start and end point group is shown.

And inputting the combined features into a vector generation model to obtain a first historical travel feature vector corresponding to the historical travel order. The vector generation model can be a convolutional neural network model, a fully-connected neural network model and the like.

The method comprises the steps of sequencing historical travel orders of each user according to the time sequence from far to near, regarding each historical travel order of each user, taking a first historical travel characteristic vector corresponding to the historical travel order and a first historical travel behavior characteristic vector of a previous historical travel order adjacent to the historical travel order in sequence as input of a characteristic vector generation layer of a travel prediction model, and predicting to obtain the first historical travel behavior characteristic vector corresponding to the historical travel order, wherein when the input first historical travel characteristic vector is input, the first historical travel behavior characteristic vector of the previous historical travel order adjacent to the historical travel order can be a preset characteristic vector.

For each history candidate start-stop point group in the history candidate start-stop point group set, a second history travel feature vector corresponding to the history candidate start-stop point group is determined based on second history travel information corresponding to the history candidate start-stop point group, for example, the second history travel information may be input to a vector generation model to obtain the second history travel feature vector.

Taking a first historical travel behavior feature vector corresponding to the historical travel order and a second historical travel feature vector corresponding to the historical candidate starting and ending point group as the input of an attention layer in a travel prediction model, and obtaining a first influence coefficient of the historical travel order on the historical candidate starting and ending point group;

and aiming at each historical travel order, calculating the product of a first influence coefficient of the historical travel order on the historical candidate starting and ending point group and the historical travel behavior characteristic vector corresponding to the historical travel order, and determining the sum of the products to obtain the first historical characteristic vector corresponding to the user.

And inputting the first historical interest area information of the user into a vector generation model to obtain a first historical interest area feature vector of the user.

And splicing the first historical characteristic vector of the user, the first historical interest region characteristic vector of the user and the historical travel behavior characteristic vector corresponding to the historical candidate starting and ending point group, and taking the spliced characteristic vector as the input of a probability prediction layer of a travel prediction model to obtain the historical probability of the user selecting the historical candidate starting and ending point group.

And selecting a history candidate start-stop point group corresponding to the maximum history probability to enable the distance between a first history trip characteristic vector corresponding to the selected history candidate start-stop point group and an actual trip characteristic vector to be minimum, and adjusting model parameters in the trip prediction model to obtain the trained trip probability prediction model.

The method uses GRU to perform user behavior representation calculation, and the specific calculation process of GRU is as follows:

z_i＝σ(x_iW_xz+h_i-1W_hz+b_z)；

u_i＝σ(x_iW_xu+h_i-1W_hu+b_u)；

W_xz，W_xu，W_xh∈R^d×h,W_hz,W_hu,W_hh∈R^h×h

wherein z is_iRepresents a reset gate vector, u_iRepresenting the update gate vector, b_zOffset vector representing reset gate, b_uOffset vector, h, representing the update gate_i-1The concealment vector representing step i-1,

representing candidate hidden vectors, b_hRepresenting the bias vectors when computing candidate concealment vectors.

Sigma is expressed as sigmoid activation function;

x_iis represented by r_iThe feature vector of (a) represents the OD of the user's ith trip;

W_xzrepresenting a first weight parameter; w_xuRepresenting a second weight parameter; w_xhRepresents a third weight parameter; w_hzRepresents a fourth weight parameter; w_huRepresents a fifth weight parameter; w_hhRepresents a sixth weight parameter; the lines indicate corresponding element multiplication, h indicates the number of hidden units, and d indicates an input size.

H is to be_iH represents the OD number of the user's historical trip as the ith trip behavior vector representation of the user Quantity, final travel behavior vector is expressed as

In addition, the invention uses an attention mechanism to dynamically capture the travel mode information of the user, and fully considers the difference influence of the historical travel record on different candidate starting and ending point groups.

After obtaining the trained travel probability prediction model, referring to fig. 3, the first planned travel information of the target user in the target time period may be determined according to the following steps:

s301, determining a candidate starting and ending point group set corresponding to the target user based on the historical travel order of the target user;

s302, inputting historical trip information of a target user and predicted trip information corresponding to the candidate start-stop point group into the trip prediction model aiming at each candidate start-stop point group in the candidate start-stop point group set, and obtaining the probability that the target user selects the candidate start-stop point group in a target time period;

s303, selecting a target starting and ending point pair from the candidate starting and ending point group set based on the probability corresponding to each candidate starting and ending point group;

s304, determining first planned travel information of the target user in a target time period according to the selected target starting and ending point pair.

When inputting the historical travel information of the target user and the predicted travel information corresponding to the candidate starting and ending point group into the travel prediction model and determining the probability that the target user selects the candidate starting and ending point in the target time period for each candidate starting and ending point group in the candidate starting and ending point group set, the method may include the following steps:

In S301, the predicted travel information includes candidate starting and ending point group information, predicted weather information, target time information, and candidate starting and ending point region distribution information, where the candidate starting and ending point region distribution information represents starting point distribution information of a region where a candidate starting point is located and ending point distribution information of a region where a candidate ending point is located, the starting point distribution information is starting point identifiers of multiple starting points distributed in the region where the candidate starting point is located, and the ending point distribution information is ending point identifiers of multiple ending points distributed in the region where the candidate ending point is located; the influence coefficient represents the influence probability degree of the historical travel order on the candidate start-stop point group selected by the user, and the larger the influence coefficient is, the larger the influence degree of the historical travel order on the candidate start-stop point group selected by the user is represented; the first historical travel information includes a travel start point and a travel end point predicted for the target user.

In a specific implementation process, the predicted trip information corresponding to the candidate starting and ending point group is input into a vector generation model, a trip characteristic vector corresponding to the candidate starting and ending point group is determined, and for each historical trip order of the target user, historical trip information corresponding to the historical trip order is input into the vector generation model, and a historical trip characteristic vector corresponding to the historical trip order is generated. The vector generation model can be a convolutional neural network model, a fully-connected neural network model and the like.

In S302, the historical travel orders of the target user are sorted according to the sequence of travel time from far to near, and for each historical travel order in the sorting, the historical travel feature vector corresponding to the historical travel order and the travel behavior feature vector corresponding to the previous historical travel order adjacent to the historical travel order in the sorting are input into a feature vector generation layer in the travel prediction model, so as to obtain the travel behavior feature vector corresponding to the historical travel order. When the historical travel feature vector of the feature vector generation layer of the input travel prediction model is the feature vector of the first historical travel order, the input travel behavior feature vector may be a vector with preset feature values all being 0.

And inputting the travel behavior characteristic vector corresponding to the historical travel order and the travel characteristic vector corresponding to the candidate starting and ending point group into an attention layer in a travel prediction model, and predicting an influence coefficient of the historical travel order on the candidate starting and ending point group when the target user selects the candidate starting and ending point group.

And aiming at each historical travel order, calculating the product of the influence coefficient of the historical travel order on the candidate starting and ending point group and the travel behavior characteristic vector corresponding to the historical travel order, calculating the sum of the products corresponding to the historical travel orders, and taking the sum as the first characteristic vector corresponding to the target user.

And inputting the first characteristic vector corresponding to the target user and the trip characteristic vector corresponding to the candidate starting and ending point group into a probability prediction layer in a trip prediction model, and predicting to obtain the probability that the target user selects the candidate starting and ending point in the target time period.

In order to improve the accuracy of obtaining the probability of selecting the candidate starting and ending point by the target user in the target time period, historical interest region information of the target user can be considered, a historical interest region feature vector for the target user is generated based on the historical interest region information of the target user, a first feature vector corresponding to the target user, a trip feature vector corresponding to the candidate starting and ending point group and the historical interest region feature vector are spliced, the spliced feature vectors are input into a probability prediction layer in a trip prediction model, the probability of selecting the candidate starting and ending point by the target user in the target time period is obtained through prediction, and the accuracy of the probability obtained through prediction is improved compared with the probability obtained only by considering the first feature vector corresponding to the target user and the trip feature vector corresponding to the candidate starting and ending point group.

Specifically, the historical interest region information may be historical interest start and end point group identifiers, historical interest start and end point region identifiers, and historical interest end point group identifiers, where the historical interest start and end point group identifiers are selected from the historical travel start and end point information, for example, the number of the same historical travel start and end point groups is determined from the historical travel orders, the historical travel start and end point groups are sorted in descending order of the number, a preset number of historical travel start and end point groups which are sorted in the top order are used as the historical interest start and end point groups, the historical interest start and end point region identifiers may be identifiers of regions where start points in the historical interest start and end point groups are located, and the historical interest end point region identifiers may be identifiers of regions where end points in the historical interest start and end points are located in the historical travel orders and may also be determined based on the historical travel start and end points in the historical travel orders.

The historical interest area information of the user is considered to be the most frequent trip area of the user on one hand and the most frequent trip starting and ending point group of the user on the other hand. The user most frequent region includes a departure place region and a destination region that the user visits most frequently in the morning, afternoon, evening.

According to the prediction method and device for the user trip starting and ending point group, all feature vectors are directly spliced and input into a model in the prior art to obtain a prediction result, only each independent feature is considered, mutual influence among the features is not considered, and the attraction of the regions with different interest point features to the user trip is changed along with time in the prior art. The method and the device fully consider the mutual influence among the characteristics and the time variability of the areas of different interest point characteristics on the user travel attraction.

In S303, the candidate start-stop point groups are sorted in descending order of probability, and a preset number of candidate start-stop point groups sorted in the top are selected as target start-stop point groups.

In S304, the target start-stop point group is used as the first planned travel information of the target time period determined for the target user, and the first planned travel information includes, in addition to the target start-stop point group, the travel distance from the target start point to the target stop point, the travel time, the traffic road condition, the weather information, and the like.

In S202, the candidate trip resources at least include trip preferential resources and trip mode resources, and the resource attribute information includes trip time information, trip distance information, trip weather information, and other information. The trip preferential resource representation is specific to the preferential degree of the planned trip information, and the trip mode resource representation is specific to the trip mode of the planned trip information, such as an office bus trip mode, a special bus trip mode, an express bus trip mode and the like.

In executing S202, selecting a target trip resource from a plurality of candidate trip resources according to the first planned trip information and resource attribute information of the candidate trip resource, may include the following steps:

Here, the travel attribute information includes information such as time information, travel distance, travel weather, travel road condition, and the like. The time information is a planned travel time, namely a target time period.

In a specific implementation process, after first planned trip information is obtained, for each candidate trip resource, comparing each piece of information in trip attribute information corresponding to the first planned trip information with each piece of information in resource attribute information of the candidate trip resource, and if at least one piece of information in trip information of the first planned trip information is matched with corresponding information in resource attribute information of the candidate trip resource, determining a target trip resource for the candidate trip resource.

For example, taking a candidate trip resource as an example, the trip attribute information of the first planned trip information includes 3 kinds of information, which are information a1, information a2 and information A3, respectively, the resource attribute information of the candidate trip resource includes 2 kinds of information, which are information B1 and information B2, respectively, and if the information a1 matches the information B2, the candidate trip resource is determined as the target trip resource.

Because the trip platform simultaneously allocates resources for a large number of users, when the candidate resources of the trip platform are limited, in order to balance global trip resources, trip resources are allocated in a balanced manner, and the planned trip information of the reference user group of the trip platform is considered, so that the target trip resources are determined for the target user, the method can comprise the following steps:

Here, the historical travel information of the reference user may refer to the historical travel information of the target user, which is not described herein again; the second planned travel information includes a set of planned starting points for the reference user, i.e., a starting point and an ending point in the travel plan.

In a specific implementation process, a reference user group is determined from a travel platform, historical travel information of each reference user in the reference user group is obtained, the historical travel information of each reference user is input into a travel prediction model, and second travel plan information of the reference user in a target time period is obtained through prediction. The process of predicting the second travel plan information of the reference user may refer to the prediction process of the first travel plan information of the target user, which is not described herein again.

For each reference user, according to the trip attribute information of the second planned trip information of the reference user and the resource attribute information of the candidate trip resource, selecting a first trip resource for the reference user from a plurality of candidate trip resources, that is, for each candidate trip resource, comparing each kind of information in the trip attribute information corresponding to the second planned trip information with each kind of information in the resource attribute information of the candidate trip resource, and if at least one kind of information in the trip information of the second planned trip information is matched with the corresponding information in the resource attribute information of the candidate trip resource, determining the candidate trip resource as the first trip resource.

When the target trip resource of the target user conflicts with the first trip resource of the reference user, for example, the first trip resource and the target trip resource are the same, and the sum of the number of the first trip resource and the number of the target trip resource exceeds the total number of the corresponding candidate trip resources in the server, the priority of the user may be considered at this time, so that the target trip resource is selected for the target user from the multiple candidate trip resources, for example, if the priority of the target user is higher than the priority of the reference user, the target trip resource is selected for the target user from the multiple candidate trip resources. The priority may be determined according to the number of completed historical travel orders of the user, and the higher the number of completed historical travel orders is, the higher the priority of the user is.

In S203, after determining the target travel resource for the target user, the target travel resource may be directly pushed to the target user, or the target travel resource may be pushed to the target user at a preset time point, which may be determined according to actual conditions.

An embodiment of the present application provides an information pushing apparatus, as shown in fig. 4, the apparatus includes:

a determining module 41, configured to input historical trip information of the target user into the trip prediction model to determine first planned trip information of the target user in the target time period;

a selecting module 42, configured to select a target trip resource from multiple candidate trip resources according to the first planned trip information and resource attribute information of the candidate trip resource;

a pushing module 43, configured to push the target trip resource to the target user.

In one embodiment, the selecting module 42 is configured to select the target travel resource according to the following steps:

In one embodiment, the determining module 41 is configured to determine the first planned travel information of the target user in the target time period according to the following steps:

In one embodiment, the determining module 41 is configured to determine the probability that the target user selects the candidate starting and ending point in the target time period according to the following steps:

acquiring historical travel orders of a reference user group;

In one embodiment, the determining module 41 is configured to determine the planned travel information of the target user in the target time period according to the following steps, including:

acquiring the road congestion degree of a target time period;

preferential resources are taken out; and (5) travel mode resources.

An embodiment of the present application further provides an electronic device 50, as shown in fig. 5, which is a schematic structural diagram of the electronic device 50 provided in the embodiment of the present application, and includes: a processor 51, a memory 52, and a bus 53. The memory 52 stores machine-readable instructions (e.g., corresponding execution instructions of the determining module 41, the selecting module 42, and the pushing module 43 in the apparatus in fig. 4, etc.) executable by the processor 51, when the electronic device 50 runs, the processor 51 communicates with the memory 52 through the bus 53, and when the processor 51 executes the following processes:

and pushing the target trip resources to the target user.

In one possible embodiment, the instructions executed by processor 51 to select a target trip resource from a plurality of candidate trip resources include:

In a possible embodiment, the instructions executed by the processor 51 for selecting a target travel resource from a plurality of candidate travel resources according to the first planned travel information and resource attribute information of the candidate travel resource include:

In one possible embodiment, the instructions executed by the processor 51 to determine the first planned travel information of the target user in the target time period include:

In one possible embodiment, the instructions executed by the processor 51 for inputting, into the travel prediction model, for each candidate starting and ending point group in the set of candidate starting and ending point groups, historical travel information of a target user and predicted travel information corresponding to the candidate starting and ending point group, and determining a probability that the target user selects the candidate starting and ending point in a target time period includes:

In one possible embodiment, the processor 51 executes instructions that determine the target time period according to the following steps:

acquiring historical travel orders of a reference user group;

In one possible embodiment, the instructions executed by the processor 51 for inputting the historical travel information of the target user into the travel prediction model to determine the planned travel information of the target user in the target time period include:

acquiring the road congestion degree of a target time period;

In a possible implementation, the processor 51 executes instructions, and the historical travel information includes any one of the following information:

In one possible embodiment, the processor 51 executes instructions, and the target trip resource includes any one of the following resources:

preferential resources are taken out; and (5) travel mode resources.

The embodiment of the present application further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the information push method are performed.

Specifically, the storage medium can be a general-purpose storage medium, such as a mobile disk, a hard disk, and the like, and when a computer program on the storage medium is executed, the information pushing method can be executed, so as to solve the problem of low utilization rate of pushed information in the prior art, the application considers historical travel information of a target user, inputs the historical travel information of the target user into a travel prediction model to obtain first planned travel information of the target user in a target time period, improves the accuracy of the obtained planned travel information to a certain extent, further selects a target travel resource for the target user from a plurality of candidate travel resources according to the first planned travel information and preset resource attribute information of the candidate travel resource, improves the matching degree between the target travel resource and the target user, and further pushes the target travel resource to the target user, the utilization rate of the pushed travel resources is improved, and the service quality of the travel platform is also improved.

It can be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the system and the apparatus described above may refer to corresponding processes in the method embodiments, and are not described in detail in this application. In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. The above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is merely a logical division, and there may be other divisions in actual implementation, and for example, a plurality of modules or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of devices or modules through some communication interfaces, and may be in an electrical, mechanical or other form.

The modules described as separate parts may or may not be physically separate, and parts displayed as modules may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.

The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing an electronic device (which may be a personal computer, a server, or a network device) to perform all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a U disk, a removable hard disk, a ROM, a RAM, a magnetic disk, or an optical disk.

The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. An information pushing apparatus, comprising:

2. The apparatus of claim 1, wherein said selection module is configured to select a target travel resource according to the following steps:

3. The apparatus of claim 1, wherein said selection module is configured to select a target travel resource according to the following steps, including:

4. The apparatus of claim 1, wherein the determining module is configured to determine the first planned travel information for the target user over the target time period according to the following steps:

5. The apparatus of claim 4, wherein the determining module is configured to determine the probability that the candidate start and end point is selected by the target user for a target time period according to:

6. The apparatus of claim 1, wherein the target time period is determined according to the following steps:

acquiring historical travel orders of a reference user group;

7. The apparatus of claim 1, wherein the determining module is configured to determine the first planned travel information for the target user over the target time period according to the following steps, including:

acquiring the road congestion degree of a target time period;

8. The apparatus of claim 1, wherein the historical travel information comprises any one of:

9. The apparatus of claim 1, wherein the target travel resource comprises any one of:

preferential resources are taken out; and (5) travel mode resources.

10. An information pushing method, characterized in that the method comprises:

and pushing the target trip resources to the target user.

11. The method of claim 10, wherein selecting a target travel resource from a plurality of candidate travel resources comprises:

12. The method of claim 10, wherein selecting a target travel resource from a plurality of candidate travel resources based on the first planned travel information and resource attribute information of the candidate travel resource comprises:

13. The method of claim 10, wherein determining the first planned travel information for the target user over the target time period comprises:

14. The method of claim 13, wherein for each candidate starting and ending point group in the set of candidate starting and ending point groups, inputting historical travel information of a target user and predicted travel information corresponding to the candidate starting and ending point group into the travel prediction model, and determining a probability that the target user selects the candidate starting and ending point in a target time period comprises:

15. The method of claim 10, wherein the target time period is determined according to the following steps:

acquiring historical travel orders of a reference user group;

16. The method of claim 10, wherein inputting the historical travel information of the target user into a travel prediction model to determine the planned travel information of the target user over the target time period comprises:

acquiring the road congestion degree of a target time period;

17. The method of claim 10, wherein the historical travel information includes any one of:

18. The method of claim 10, wherein said target travel resources comprise any one of the following:

preferential resources are taken out; and (5) travel mode resources.

19. An electronic device, comprising: a processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, when an electronic device runs, the processor and the storage medium communicate through the bus, and the processor executes the machine-readable instructions to execute the steps of the information pushing method according to any one of claims 10 to 18.

20. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, performs the steps of the information push method according to any one of claims 10 to 18.