CN111854781A

CN111854781A - Navigation path recommendation method and device and electronic equipment

Info

Publication number: CN111854781A
Application number: CN202010555341.0A
Authority: CN
Inventors: 张凌宇; 李丹
Original assignee: Beijing Didi Infinity Technology and Development Co Ltd
Current assignee: Beijing Didi Infinity Technology and Development Co Ltd
Priority date: 2020-06-17
Filing date: 2020-06-17
Publication date: 2020-10-30

Abstract

The application provides a navigation path recommendation method, a navigation path recommendation device and electronic equipment, wherein the method comprises the following steps: obtaining a cluster map generated based on a plurality of first objects fed back over selection bias; the cluster map comprises a plurality of cluster subgraphs and stroke mode selection probabilities corresponding to the cluster subgraphs; determining a target subgraph associated with a target object from the cluster subgraphs; and recommending and sequencing available paths according to the travel mode selection probability corresponding to the target subgraph. The path recommendation is more in line with the expectation of the target object, and the use experience of the target object is improved while the intelligent degree of the path recommendation is improved.

Description

Navigation path recommendation method and device and electronic equipment

Technical Field

The application relates to the technical field of navigation, in particular to a navigation path recommendation method and device and electronic equipment.

Background

Currently, navigation software can plan different travel routes for a user based on different travel modes for the user to select. However, no matter how many travel routes can be planned, the end user only chooses one to use according to the preference. Obviously, too many travel routes arranged in disorder are provided for the user, so that the intelligence of the navigation service is insufficient, and the time consumption for selecting the travel routes by the user is increased, and the use experience of the user is influenced.

Disclosure of Invention

In view of the above, an object of the present application is to provide a navigation path recommendation method, a navigation path recommendation apparatus, and an electronic device, which are capable of recommending a target object according to the degree that each analyzed available path is close to the expectation of the target object by analyzing the preference of the target object for selecting a route. Therefore, the problem that time consumption is too long when a user searches for a path which accords with preference from available paths in the prior art is solved, the effect of improving the recommendation accuracy and recommendation efficiency of the available paths is achieved, and the use experience of the user is improved.

In a first aspect, an embodiment of the present invention provides a navigation path recommendation method, including:

acquiring a cluster map generated based on a plurality of first objects; the first object is a service request object which feeds back selection preference; the cluster map comprises a plurality of cluster subgraphs and stroke mode selection probabilities corresponding to the cluster subgraphs; the travel mode selection probability is the probability that the first object belonging to the clustering subgraph selects the path planned by each type of travel mode;

determining a target subgraph associated with a target object from the cluster subgraphs;

recommending and sequencing available paths according to the travel mode selection probability corresponding to the target subgraph; wherein the available paths include paths that are planned between a start point and an end point specified by the target object using different ones of the travel patterns.

In an alternative embodiment, the step of generating a cluster map based on a plurality of first objects comprises:

acquiring characteristic information of the first object on multiple dimensions;

mapping the first object to an initial cluster map according to the characteristic information on the multiple dimensions, and constructing edge weights among different first objects;

cutting the initial clustering graph according to the edge weight, so that the edge weight belonging to different clustering subgraphs is smaller than the edge weight belonging to the same clustering subgraph;

counting the travel mode selection probability corresponding to each type of travel mode based on historical travel data of the first object belonging to the clustering subgraph;

and establishing a corresponding relation between the clustering subgraph and the travel mode selection probability to obtain the clustering graph.

In an optional embodiment, the step of acquiring feature information of the first object in multiple dimensions includes:

obtaining a plurality of user representation data of a specified type associated with the first object;

and preprocessing the user portrait data according to the acquired data missing proportion of the user portrait data to obtain the characteristic information on multiple dimensions.

In an alternative embodiment, the step of preprocessing the user image data comprises:

and selecting to delete the user portrait data of the first object or perform characteristic engineering processing on the user portrait data of the first object according to the data missing proportion of each first object.

In an optional embodiment, the mapping the first object into an initial cluster map according to the feature information on the multiple dimensions, and constructing edge weights between different first objects includes:

constructing a first feature vector according to the feature information of each first object on multiple dimensions;

determining the corresponding positions of the first objects on the initial cluster map according to the first feature vectors, so that the distance between different first objects on the initial cluster map is the vector difference between the corresponding first feature vectors;

calculating the corresponding edge weight according to a vector difference between the first feature vectors of the different first objects.

In an optional embodiment, the step of determining a target subgraph associated with a target object from the clustered subgraphs comprises:

Acquiring characteristic information of the target object on multiple dimensions, and constructing a corresponding second characteristic vector;

calculating a vector difference between the second feature vector and a third feature vector corresponding to a clustering center of each clustering subgraph;

and determining the clustering subgraph corresponding to the third feature vector with the minimum vector difference with the second feature vector as the target subgraph.

In an optional embodiment, the step of performing recommendation ranking on available paths according to the travel mode selection probability corresponding to the target sub-graph includes:

sorting the available paths according to the corresponding travel mode selection probability and the sequence from big to small;

and recommending the available paths according to the sequencing sequence.

In an alternative embodiment, the method further comprises:

acquiring a target path selected by the target object from the available paths;

and updating the target sub-graph and the travel mode selection probability corresponding to the target sub-graph according to the target object and the travel mode corresponding to the target path.

In an optional embodiment, the step of obtaining a target path used by the target object to select from the available paths further includes:

And when the target object feeds back the selection deviation based on the selected target path, acquiring the target path.

In a second aspect, an embodiment of the present invention provides a navigation path recommendation apparatus, including:

an obtaining module, configured to obtain a cluster map generated based on a plurality of first objects; the first object is a service request object which feeds back selection preference; the cluster map comprises a plurality of cluster subgraphs and stroke mode selection probabilities corresponding to the cluster subgraphs; the travel mode selection probability is the probability that the first object belonging to the clustering subgraph selects the path planned by each type of travel mode;

the determining module is used for determining a target subgraph associated with a target object from the clustering subgraphs;

the recommending module is used for recommending and sequencing the available paths according to the travel mode selection probability corresponding to the target subgraph; wherein the available paths include paths that are planned between a start point and an end point specified by the target object using different ones of the travel patterns.

In an alternative embodiment, the apparatus comprises:

the extraction module is used for acquiring multi-dimensional characteristic information of the first object;

The building module is used for mapping the first object to an initial cluster map according to the characteristic information on the multiple dimensions and building edge weights among different first objects;

the cutting module is used for cutting the initial clustering graph according to the edge weight so that the edge weight belonging to different clustering subgraphs is smaller than the edge weight belonging to the same clustering subgraph;

a counting module, configured to count the travel pattern selection probability of each type of the travel pattern based on historical travel data of the first object belonging to the clustered sub-graph;

and the establishing module is used for establishing the corresponding relation between the clustering subgraph and the travel mode selection probability so as to obtain the clustering graph.

In an alternative embodiment, the apparatus further comprises:

the acquisition module is further configured to acquire a target path selected by the target object from the available paths;

and the updating module is used for updating the target sub-graph and the stroke mode selection probability corresponding to the target sub-graph according to the target object and the stroke mode corresponding to the target path.

In an optional embodiment, the obtaining module is specifically configured to: and when the target object feeds back the selection deviation based on the selected target path, acquiring the target path.

In a third aspect, an embodiment of the present invention provides an electronic device, including: the system 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 is communicated with the storage medium through the bus, and the processor executes the machine-readable instructions to execute the steps of the method according to any one of the preceding implementation modes.

In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, where the computer program is executed by a processor to perform the steps of the method according to any one of the foregoing embodiments.

Based on any one of the above aspects, the navigation path recommendation method provided by the embodiment of the present invention obtains the cluster map generated based on the plurality of first objects, where the clustered plurality of first objects are all service request objects for which preference is fed back, and by selecting the probability of the travel pattern corresponding to the cluster subgraph representing different classes of first objects in the cluster map, the preference of different types of objects in selecting an available path can be analyzed. And then, determining a target subgraph associated with the target object from the clustered subgraphs so as to identify the type of the target object, and recommending and sequencing available paths planned by different travel modes according to the travel mode selection probability corresponding to the target subgraph. Therefore, the method and the device realize the sequencing of the available paths based on the analyzed selection preference of the target object, facilitate the target object to quickly obtain the path close to the selection preference of the target object, and shorten the time for selecting the route by a user. The recommendation is more in line with the expectation of the target object, the intelligent degree of path recommendation is improved, and the use experience of the target object 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 is a schematic diagram illustrating an architecture of a navigation path recommendation system according to an embodiment of the present application;

FIG. 2 is a flowchart illustrating a navigation path recommendation method according to an embodiment of the present disclosure;

fig. 3 illustrates a second flowchart of a navigation path recommendation method according to an embodiment of the present application;

fig. 4 is a flowchart illustrating sub-steps of step S202 in a navigation path recommendation method according to an embodiment of the present application shown in fig. 3;

FIG. 5 is a diagram illustrating an example of a clustering result provided by an embodiment of the present application;

fig. 6 is a flowchart illustrating sub-steps of step S102 in a navigation path recommendation method according to an embodiment of the present application shown in fig. 2;

FIG. 7 is a third flowchart illustrating a navigation path recommendation method according to an embodiment of the present application;

Fig. 8 is a schematic diagram illustrating a navigation path recommendation apparatus according to 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.

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 "user," "object," "service request object," "service requestor," and "customer" are used interchangeably in this application to refer to an individual, entity, or tool that can request or subscribe to a service. The term "target object" in the present application may also be converted into "first object" in some cases.

The Positioning technology used in the present application may be based on a Global Positioning System (GPS), a Global Navigation Satellite System (GLONASS), a COMPASS Navigation System (COMPASS), a galileo Positioning System, a Quasi-Zenith Satellite System (QZSS), a Wireless Fidelity (WiFi) Positioning technology, or the like, or any combination thereof. One or more of the above-described positioning systems may be used interchangeably in this application.

It is noted that before the application is filed, various routes are planned by using travel modes with different advantages for a user to select, and the goal of the navigation service is pursued. For example, the route planning is performed based on the travel mode with the shortest required time, the travel mode with the fastest transfer, the travel mode using a shared bicycle, the travel mode with the fewest walks, the travel mode using public transportation, the travel mode using an automobile, and the like, so that the diversity of available routes is ensured for the user to select. However, the diversified paths complete the user's selectable space, but the user needs more time to find a path that meets the preference, which is only complete but not intelligent enough for the whole navigation service.

Of course, in some related technologies, an attempt is also made to evaluate the preference of the user in selecting the travel mode by sharing the travel history of the user. However, the above-described approach is only effective for users who explicitly feed back selection preferences or who have fed back approval for the route used. In fact, the users who have not explicitly fed back the selection preference or who have fed back approval to the used route account for the majority of the user group, obviously, the methods used in the related art cannot make accurate analysis on their preferences, and even cannot make accurate and personalized intelligent recommendation on their basis.

Therefore, the navigation path recommendation method provided by the application can analyze the path selection deviation of most crowds by feeding back a small part of the crowds with the path selection deviation, and further realize intelligent personalized recommendation of all users.

Fig. 1 is a schematic architecture diagram of a navigation path recommendation system 100 according to an embodiment of the present disclosure. For example, the navigation path recommendation system 100 may be a mobile travel platform that aggregates transportation services, such as taxis, designated driving services, express, carpools, bus services, driver rentals, or regular service, or any combination thereof, that may plan diverse paths based on different travel patterns between a designated start point and end point. The navigation path recommendation system 100 may include one or more of a server 110, a network 120, a service requester 130, 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 starting point and the ending point for which path planning is required based on the service request obtained from the service requester 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 (S)). 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, RISC), a microprocessor, or the like, or any combination thereof.

In some embodiments, the device type corresponding to the service request end 130 may be a mobile device, such as a smart home device, a wearable device, a smart mobile device, a virtual reality device, an augmented reality device, or the like, or may be a tablet computer, a laptop computer, or a built-in device in a motor vehicle, or the like.

The service request terminal 130 may be controlled by a service request object to feed back service requirements to the server 110. It is understood that the service requester 130 and the service request object have a relatively fixed binding relationship, and the server 110 may confirm the service request object actually requesting the navigation service according to the service requester 130 sending the request. The user profile data of the service request object may be transmitted to the server 110 via the service requester 130.

For convenience of description, in the embodiment of the present invention, the service request end 130 and the service request object may be used interchangeably.

In some embodiments, a database 150 may be connected to the network 120 to communicate with one or more components (e.g., the server 110, the service requester 130, the service provider 140, etc.) in the navigation path recommendation system 100. One or more components in the navigation path recommendation system 100 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 navigation path recommendation system 100, or the database 150 may be part of the server 110. In some embodiments, the database 150 may store user representation data corresponding to all service requestors 130.

The following describes in detail a navigation path recommendation method provided in an embodiment of the present application with reference to the content described in the navigation path recommendation system 100 shown in fig. 1.

Referring to fig. 2, a flowchart of a navigation path recommendation method provided in the embodiment of the present application is schematically shown, where the method may be executed by a server in the navigation path recommendation system 100, and the specific execution process includes:

in step S101, a cluster map generated based on a plurality of first objects is acquired.

The first object is a service request object which feeds back the preference of selection. The selection bias may characterize the preference of the service request object to select a travel mode. It can be understood that the service request object fed back with the selection preference usually performs the routing according to the selection preference of the service request object when the navigation service is used at ordinary times.

The cluster map may be used to present clustering results for a plurality of first objects. The clustering result includes a plurality of cluster clusters, each cluster being clustered by first objects that are similar in a specified direction. Correspondingly, the cluster clusters are characterized in the cluster map through a plurality of cluster subgraphs. The first object is presented in the form of a point in the cluster map, and the point is also associated with the related information of the first object. And the first objects represented by the points in the same clustering subgraph belong to the same clustering cluster.

And step S102, determining a target subgraph associated with the target object from the clustered subgraphs.

The target object may also be a service request object for which the server needs to provide navigation services. It should be noted that, on the one hand, there is a difference between the first object and the target object, that is, the target object is an object that the server needs to provide a service at the time, and the first object is an object that the server has previously provided a service and has fed back a preference. On the other hand, there is no absolute boundary between the first object and the target object, and the target object may also become the first object under some conditions, for example, when the target object uses the navigation service and feeds back the selection bias to the server, the target object is also divided into the first object.

The target subgraph is a clustering subgraph associated with a target object. The association may be expressed as that the target object may be clustered into a cluster characterized by the target subgraph. In other words, the target object has a certain commonality with the first object in the cluster corresponding to the target sub-graph, and the selection bias of the first object corresponding to the target sub-graph also has a commonality with the target object.

And step S103, recommending and sequencing the available paths according to the travel mode selection probability corresponding to the target subgraph.

The available paths may include paths drawn between a start point and an end point specified by the target object using different travel patterns.

The travel modes can be understood as strategies adopted for planning the paths, and different travel modes correspond to different advantages or cooperation among different travel tools, so that diversified and high-quality path sets can be provided under diversified travel modes. For example, the journey mode may include a policy that minimizes static paths, a policy that minimizes required time, a policy that plans paths based primarily on public transportation, a policy that plans minimal walks, a policy that achieves minimal transfers, a policy that plans paths based on vehicle driving, and so forth.

The travel pattern selection probabilities described above may be used to characterize a population's preferences in selecting navigation routes. It is understood that each cluster subgraph in the cluster map can characterize a population obtained by classifying the first object. Therefore, each clustering subgraph is also established with the corresponding relation with the travel mode selection probability. The travel mode selection probability is the probability that the planned path of each type of travel mode is selected by a group consisting of the first objects belonging to the clustering subgraphs.

In the embodiment of the invention, the available paths are sequenced based on the travel mode selection probability corresponding to the target subgraph, so that the available paths are sequenced according to the sequence closer to the travel mode selection deviation of the target object, the target object can conveniently and quickly lock the available paths meeting the expectation, even the available paths meeting the expectation of the target object can be directly recommended to the target object, the selection time cost brought by the diversified and high-quality available path set is saved, and the use experience of the target object is also improved.

Therefore, the embodiment of the invention analyzes and processes the service request object with small magnitude and applies the analysis result to the service request object with large range. Therefore, even if the service request object with the preference is not fed back, the available path can be accurately recommended to the service request object, personalized service is realized, and user experience is improved.

Specific details of embodiments of the invention are described below:

the purpose of the above step S101 is to obtain a cluster analysis result of a first object of a small order of magnitude. In some embodiments, the cluster map that has been generated may be obtained directly. Therefore, as shown in fig. 3 on the basis of fig. 2, the above method may include the following steps:

In step S201, feature information of the first object in multiple dimensions is acquired.

In some embodiments, the feature information of the first object in multiple dimensions may be description data of the first object. Optionally, the feature information in multiple dimensions may not only characterize the first object, but may also be used to analyze similarities between different first objects, thereby facilitating clustering.

In some embodiments, the server stores user representation data for the service request object. It should be understood that the user portrait data is raw data for describing the service request object, and may describe the service request object from various aspects such as personal information, living environment information, and travel information. Therefore, the step S201 may include: a plurality of user representation data of a specified type associated with the first object is queried. For example, the user imaging data may include age, gender, city class, order quantity, whether to make an individual, business trip probability, visitor probability, and the like.

In other embodiments, the user pictorial data of the service request object may be stored on a third party platform. Therefore, the step S201 may include: a plurality of user representation data of a specified type associated with a first object is obtained from a third party platform.

There are generally few service request objects that feed back to the server their true selection bias. That is, the amount of data of the first object is small, and the related data of each first object is precious. Meanwhile, due to objective reasons, the acquired user portrait data corresponding to the first object is incomplete (a null value or an error value exists), and the incomplete data is not only unfavorable for analysis, but also affects the accuracy of an analysis result. Therefore, in order to guarantee the accuracy of the analysis and make the most of the available data, the step S201 may further include, after acquiring the user portrait data related to the first object: and preprocessing the user image data according to the data missing proportion of the acquired user image data.

In one embodiment, the preprocessing the user image data according to the data loss ratio of the acquired user image data may include: and selecting to delete the user portrait data of the first object or perform characteristic engineering processing on the user portrait data of the first object according to the data missing proportion of the user portrait data corresponding to each first object. For example, if the data loss ratio of the user image data corresponding to a first object exceeds a first threshold, the user image data corresponding to the first object is deleted. For another example, if the data missing ratio of the user portrait data corresponding to a first object is lower than the second threshold, the user portrait data of the first object is processed by feature engineering.

It will be appreciated that the first threshold mentioned in the above examples is greater than the second threshold. The feature engineering process may be, but is not limited to, feature missing value filling, One-Hot Encoding (One-Hot Encoding) on discrete features, normalization on continuous features, balancing positive and negative sample proportions, and the like.

In other embodiments, in order to better utilize the limited data, the distribution of the missing data in the corresponding user portrait data may be analyzed after the missing data ratio exceeds a specific value (e.g., a first threshold), and according to the distribution, the user portrait data is selected to be deleted or supplemented. It will be appreciated that user representation data is data describing the service request object from multiple perspectives, i.e. the user representation data includes description information in different directions.

Therefore, if the distribution of the missing data indicates that the data missing amount corresponding to each missing position point is smaller than the specified value, and the missing position points are dispersed in the description information in multiple directions, inference (for example, using a statistical reference) can be performed based on other information in the user portrait data to complete the supplement of the missing data. For example, if the user image data lacks age information but has identification card information, the age information can be derived from the identification card information.

After the user image data is preprocessed, multi-dimensional characteristic information can be extracted based on the user image data.

In some embodiments, the plurality of dimensions may be specified dimensions, and may include, for example, an age dimension, a gender dimension, a city dimension, a service request number dimension, whether to be singleton, a business travel probability dimension, a guest probability dimension, and the like. In this way, the relevant feature information is extracted based on the actual information of the first object corresponding to the plurality of dimensions.

For convenience of processing, the selectable items corresponding to each dimension may be digitized, that is, different selectable items in the same dimension are represented by different numbers. For example, the gender dimension includes two options, male and female, and after digital processing, female is characterized by "0" and male is characterized by "1". Thus, when the gender of the first object is female, the characteristic information of the first object extracted from the gender dimension is 0.

Step S202, according to the characteristic information on the multi-dimension, the first object is mapped to the initial cluster map, and the edge weight between different first objects is constructed.

In some embodiments, as shown in fig. 4, the step S202 may include:

s202-1, constructing a first feature vector according to the feature information of each first object on multiple dimensions.

In some embodiments, the feature information of the first objects in each dimension may be normalized, and the normalized feature information corresponding to each first object may be arranged according to the same order to obtain a first feature vector.

S202-2, determining the corresponding positions of the first objects on the initial cluster map according to the first feature vectors, so that the distance between different first objects on the initial cluster map is the vector difference between the corresponding first feature vectors.

Different points in the initial cluster map correspond to different feature vectors. The distance between different points also represents the vector difference between the corresponding feature vectors. Based on the first feature vectors of the first objects, the first objects are mapped onto the initial cluster map such that the distance of different first objects on the initial cluster map may characterize the vector difference between the corresponding first feature vectors.

For example, the effect of mapping the first object to the initial cluster map is shown in FIG. 5.

S202-3, corresponding edge weights are calculated according to vector differences between the first feature vectors of different first objects.

It can be understood that the edge weight of the first object on the initial cluster map corresponding to the two first feature vectors with smaller vector difference is larger, and vice versa.

Step S203, the initial cluster map is cut according to the edge weight.

In some embodiments, the number of clustering centers capable of obtaining the best clustering effect may be determined, and then the clustering sub-graphs are partitioned according to the edge weights corresponding to different first objects, so as to partition the clustering sub-graphs with the same number as the clustering centers. In addition, when clustering subgraph division is carried out, the principle is that the weight of the edge belonging to different clustering subgraphs is smaller than the weight of the edge belonging to the same clustering subgraph.

In some embodiments, the method for determining the number of the clustering centers with the best clustering effect may be to successively try to perform clustering by using the number of the clustering centers with different values, and then evaluate the clustering effect of clustering by using different numbers of the clustering centers by using Calinski-Harabasz Score, so as to determine the number of the clustering centers with the best clustering effect.

It should be noted that, after the clustering subgraph is cut out, the clustering center of the clustering subgraph can be calculated according to all the first feature vectors belonging to the clustering subgraph. The cluster center also corresponds to a feature vector.

And step S204, counting the travel mode selection probability corresponding to each type of travel mode based on the historical travel data of the first object belonging to the clustering subgraph.

The historical travel data is historical record data of the navigation service used by the first object, and may include an available path provided by the navigation service request of the server for the first object, a travel mode used for planning the available path, and a travel mode corresponding to the actually used available path.

In some embodiments, all historical travel data corresponding to the first object belonging to the same cluster sub-graph may be obtained, and the probability that routes planned by different travel modes are selected is counted as the travel mode selection probability corresponding to the cluster sub-graph.

In other embodiments, the travel data carried by all the first objects belonging to the same clustering subgraph when the feedback selection is biased can be obtained, and the probability that routes planned by different travel modes are selected is counted to be used as the travel mode selection probability corresponding to the clustering subgraph.

It is understood that the preference of the service request object feedback selection is mostly based on the feedback of the one-time navigation service. For example, in the process that the service request object requests a navigation service once, if screenshot or sharing is performed based on the selected available path, it can be determined that the service request object is satisfied or not satisfied with the used available path, so as to determine the selection bias of the service request object.

Of course, the aforementioned "sharing" operations may generally characterize the satisfaction of the service request object with the available paths. And the "screenshot" operation may analyze whether the service request object is satisfied with the available path to use based on subsequent other operations. For example, after the screenshot operation, the service request object triggers a wrong path or reports a path, and the like, and it can be determined that the service request object is unsatisfied with the available path, and the other objects represent satisfaction.

In addition, in other embodiments, only the historical trip data that the first object belonging to the same clustering subgraph is satisfied with the available paths may be counted. Or when the historical travel data of the first object belonging to the same clustering subgraph are counted, the travel mode selection probability corresponding to the clustering subgraph is calibrated by using the occurrence frequency of the travel mode unsatisfactory to the first object.

Based on the above, in some embodiments, the determining of the first object includes: and determining the service object as a first object when detecting that the service request object shares the historical travel data through screenshot or triggering a sharing key.

And step S205, establishing a corresponding relation between the clustering subgraphs and the travel mode selection probability to obtain a clustering graph.

In the above embodiment, the plurality of dimensions mentioned in step S201 may be specified. Of course, in other embodiments, all dimensions that may be selected may also be predetermined. Then, different dimensions are combined, and feature information corresponding to each first object under different dimension combinations is respectively extracted. And clustering the first objects respectively based on the feature information extracted under different dimension combinations. And finally, evaluating the clustering effect of clustering by adopting the characteristic information under different dimension combinations. Therefore, the dimension combination with the best clustering effect can be determined, and the dimension included in the dimension combination is used as the dimension for extracting the characteristic information in the actual application process.

For example, all dimensions that can be selected include A, B, C, then A, B are combined, A, C are combined and B, C is combined. And extracting characteristic information of the first object from the two dimensions AB, and clustering to obtain a clustering result A. And extracting characteristic information of the first object from two dimensions of the AC, and clustering to obtain a clustering result B. And extracting characteristic information of the first object from two dimensions of BC, and clustering to obtain a clustering result C. And respectively evaluating the clustering result A, the clustering result B and the clustering result C, and if the evaluation determines that the clustering effect of the clustering result A is optimal, taking the two dimensions AB as the plurality of dimensions mentioned in the step S201.

In some other embodiments, the step S101 may also be a process of generating a cluster map. In this embodiment, the implementation process of step S101 may refer to steps S201 to S205, which are not described herein again.

The purpose of step S102 is to identify a group to which a service request object including no preference feedback belongs, and to infer the preference of most service request objects based on the principle that the same group has a certain commonality in the selection trip mode.

In some embodiments, the target subgraph to which the target object belongs can be determined by evaluating the correlation between the target object and each clustered subgraph.

In some embodiments, as shown in fig. 6, the step S102 may include the following sub-steps:

and a substep S102-1, acquiring characteristic information of the target object on multiple dimensions, and constructing a corresponding second characteristic vector.

In some embodiments, the plurality of dimensions are the same as those mentioned in step S201. Acquiring characteristic information of a plurality of dimensions is also the same as the principle of step S201, except that the targeted object is changed from the first object to the target object. Therefore, the description thereof is omitted. It should be noted that, when constructing the second feature vector, the arrangement order of the feature information in different dimensions is the same as that of the first feature vector.

And a substep S102-2, calculating a vector difference between the second feature vector and a third feature vector corresponding to the clustering center of each clustering subgraph.

In some embodiments, an edge weight between the target object and the cluster center of each clustered subgraph may also be determined.

And a substep S102-3, determining the clustering subgraph corresponding to the third characteristic vector with the minimum vector difference with the second characteristic vector as the target subgraph.

In some embodiments, the clustering subgraph to which the clustering center with the largest edge weight with the target object belongs may be used as the target subgraph.

In some embodiments, when a plurality of target objects are faced and none of the target objects has a feedback selection bias, the plurality of target objects may be clustered to obtain a plurality of target clustering subgraphs. And comparing and analyzing the cluster subgraph corresponding to the first object with the target cluster subgraph, and determining a target subgraph associated with each target cluster subgraph from the cluster subgraph corresponding to the first object, thereby obtaining a target subgraph corresponding to a target object belonging to each target cluster subgraph.

The analysis mode between the clustering subgraphs and the target clustering subgraphs can be to calculate the weight edges between the target clustering subgraphs and the clustering centers of the clustering subgraphs. The target subgraph can be determined by determining the cluster subgraph to which the cluster center with the largest weight edge with the target cluster subgraph belongs as the target subgraph.

In some embodiments, the step S103 is to sequentially present available paths to the target object, so as to implement targeted intelligent recommendation.

In some embodiments, in step S103, the available paths may be sorted in descending order according to the corresponding trip mode selection probability, and the available paths are recommended according to the sorting order.

To facilitate understanding of the present application by those skilled in the art, the following is an example:

user portrait data of a service request object (namely a first object) having a sharing or screenshot operation in a specified time period is acquired, and feature information on multiple dimensions such as gender, age, city level, number of orders for vehicle transfer, whether the user is single, business travel probability, tourist probability and the like is extracted. In the gender dimension, female is represented by 0 and male is represented by 1. For the city scale dimension: the first-line city, the second-line city, and the third-four-line city are denoted by 1, 2, and 3, respectively, and then normalized. Whether to take values of 0 and 1 on its own. The business travel probability, the tourist probability and the consumption capacity value are all in the interval (0, 1).

And clustering the extracted first object according to the 5 clustering centers to obtain 5 clustering subgraphs. Each cluster center or each cluster subgraph corresponds to a cluster. The clustering results obtained are shown in the following table:

The characteristic information of the cluster centers corresponding to different cluster clusters is shown in the table above. Then, the historical travel data of the first objects included in each cluster is counted to obtain the travel mode selection probability as follows:

by combining the two tables, it can be easily known that the third eigenvector corresponding to the cluster center of the cluster subgraph corresponding to the cluster sub-graph with the cluster number of 0 is (0,0.317,1,0.09,0.958,0.269,0.259), the probability of selecting the fastest mode of the cluster sub-graph with the cluster number of 0 is 0.15, the probability of selecting the fewest walking mode is 0.1, the probability of selecting the fewest transfer mode is 0.12, the probability of selecting the no-subway mode is 0.35, the probability of selecting the recommended mode is 0.92, the probability of selecting to start the shared bicycle is 0.2, the probability of selecting to start the regular bus/express line is 0.63, and the probability of selecting to start the taxi taking is 0.09. And the others are analogized in turn.

And then, evaluating a cluster to which the target object belongs according to the third eigenvector corresponding to each cluster coarseness and the second eigenvector corresponding to the target object, and taking a cluster subgraph corresponding to the cluster as a target subgraph. And then the corresponding travel mode selection probability is found through the table. And finally, recommending and sequencing the available paths planned by different travel modes according to the travel mode selection probability, and presenting the available paths to the target object. And personalized intelligent recommendation is realized.

In some embodiments, as shown in fig. 7, the navigation path recommendation method may further include the following steps:

in step S301, a target object is acquired to select a target path to be used from available paths.

In some embodiments, after each target object has used up the navigation service, the target path selected by the target object in the navigation service at that time may be read.

In some other embodiments, the target path may be acquired when the target object feeds back the selection bias based on the selected target path. For example, after the target object is subjected to screenshot or sharing, a target path selected by the target object is obtained.

Step S302, updating the target sub-graph and the stroke mode selection probability corresponding to the target sub-graph according to the stroke modes corresponding to the target object and the target path.

In some embodiments, the cluster center of the target sub-graph may be updated according to the target object, and the trip mode selection probability corresponding to the target sub-graph is updated by using the current trip data.

Based on the same inventive concept, a navigation path recommendation device corresponding to the navigation path recommendation method is further provided in the embodiments of the present application, and as the principle of solving the problem of the device in the embodiments of the present application is similar to that of the navigation path recommendation method in the embodiments of the present application, the implementation of the device may refer to the implementation of the method, and repeated details are not repeated.

Referring to fig. 8, which is a schematic view of a navigation path recommendation device provided in the fifth embodiment of the present application, the device includes: the device comprises an acquisition module, a determination module and a recommendation module; wherein:

an obtaining module, configured to obtain a cluster map generated based on a plurality of first objects; the first object is a service request object which feeds back selection preference; the cluster map comprises a plurality of cluster subgraphs and stroke mode selection probabilities corresponding to the cluster subgraphs; and the travel mode selection probability is the probability that the first object belonging to the clustering subgraph selects the planned path of each type of travel mode.

And the determining module is used for determining a target subgraph associated with the target object from the clustering subgraphs.

Optionally, the apparatus may further include:

and the extraction module is used for acquiring the characteristic information of the first object on multiple dimensions.

And the construction module is used for mapping the first object to an initial cluster map according to the characteristic information on the multiple dimensions and constructing edge weights among different first objects.

And the cutting module is used for cutting the initial clustering graph according to the edge weight so as to enable the edge weight belonging to different clustering subgraphs to be smaller than the edge weight belonging to the same clustering subgraph.

And the counting module is used for counting the travel mode selection probability of each type of travel mode based on the historical travel data of the first object belonging to the clustering subgraph.

Optionally, the extracting module specifically includes:

Optionally, the extracting module is further configured to select to delete the user portrait data of the first object or perform feature engineering processing on the user portrait data of the first object according to the data missing proportion of each first object.

Optionally, the building module is specifically configured to:

Optionally, the determining module is specifically configured to:

acquiring the characteristic information of the target object on multiple dimensions, and constructing a corresponding second characteristic vector;

Optionally, the recommendation module is specifically configured to:

and recommending the available paths according to the sequencing sequence.

Optionally, the apparatus further comprises:

Optionally, the obtaining module is specifically configured to: and when the target object feeds back the selection deviation based on the selected target path, acquiring the target path.

An 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 navigation path recommendation method are performed.

Specifically, the storage medium can be a general storage medium, such as a mobile disk, a hard disk, and the like, and when a computer program on the storage medium is run, the navigation path recommendation method can be executed, so that the problem of long time consumption for searching a path meeting the preference from available paths is solved, the recommendation accuracy and recommendation efficiency of the available paths are improved, and the use experience of a user is 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, in order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, functional units in various embodiments of the present application may be integrated into one body, and the technical solution 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. 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.

Claims

1. A navigation path recommendation method is characterized by comprising the following steps:

2. The method of claim 1, wherein the step of generating the cluster map based on the first plurality of objects comprises:

3. The method of claim 2, wherein the step of obtaining feature information of the first object in multiple dimensions comprises:

4. The method of claim 3, wherein the step of pre-processing the user imagery data comprises:

and selecting to delete the user portrait data of the first object or perform feature engineering processing on the user portrait data of the first object according to the data missing proportion corresponding to each first object.

5. The method of claim 2, wherein the step of mapping the first objects into an initial cluster map according to the multi-dimensional feature information and constructing edge weights between different first objects comprises:

determining the corresponding positions of the first objects on the initial cluster map according to the first feature vectors, so that the distances of different first objects on the initial cluster map are vector differences between the corresponding first feature vectors;

6. The method of claim 5, wherein the step of determining a target subgraph associated with the target object from the clustered subgraphs comprises:

7. The method of claim 1, wherein the step of performing a recommended ordering of available paths according to the travel pattern selection probability corresponding to the target sub-graph comprises:

and recommending the available paths according to the sequencing sequence.

8. The method of claim 1, further comprising:

Acquiring a target path selected by the target object from the available paths;

9. The method of claim 8, wherein the step of obtaining the target path used by the target object to select from the available paths further comprises:

10. A navigation path recommendation device, comprising:

11. The apparatus of claim 10, further comprising:

12. The apparatus of claim 10, further comprising:

13. An electronic device, comprising: a processor, a storage medium and a bus, the storage medium storing machine-readable instructions executable by the processor, the processor and the storage medium communicating via the bus when the electronic device is operating, the processor executing the machine-readable instructions to perform the steps of the method according to any one of claims 1 to 9.

14. A computer-readable storage medium, having stored thereon a computer program which, when being executed by a processor, is adapted to carry out the steps of the method according to any one of claims 1 to 9.