CN110008413B - Traffic travel problem query method and device - Google Patents

Abstract

The embodiment of the application relates to the technical field of traffic, in particular to a traffic travel problem query method and device, which are used for improving the accuracy of identifying user intention and reducing man-machine interaction times. Receiving traffic information to be queried; identifying a query intention corresponding to traffic information to be queried; extracting first descriptive information corresponding to the descriptive information type from traffic information to be queried according to the descriptive information type corresponding to the query intention, wherein the first descriptive information comprises position descriptive information and relation descriptive information; searching a starting point entity corresponding to the position description information from the traffic knowledge graph; the traffic knowledge graph comprises entities of each layer in a traffic network and connection relations among the entities; determining a target entity according to the relation description information, the starting point entity and the connection relation among the entities included in the traffic knowledge graph; and reasoning and inquiring a target result corresponding to the intention according to the traffic knowledge graph, the starting point entity and the target entity. Therefore, the accuracy of identifying the user intention can be improved, and the man-machine interaction times are reduced.

Description

Traffic travel problem query method and device

Technical Field

The embodiment of the application relates to the technical field of traffic, in particular to a traffic travel problem query method and device.

Background

In the electronic map commonly used in the electronic equipment at present, the map search engine organizes data in a way of position points on the map, but no relation between any two position points is described. Therefore, when using the existing inquiry function of the electronic map, the user needs to input a designated start point, an end point and a travel mode to plan a route. This query function can only be applied to simple question scenes entered by the user. In addition to the simple problem scenario described above, in some other cases, a scenario may be encountered in which a user inputs a complex problem, such as a query for a blurred geographic location scenario, a query for a route transfer scenario, a query for a nearby range of locations, and the like, the complex problem scenario includes, but is not limited to, several of the above, and various combinations of any of the above.

In the conventional electronic map, the conventional map search engine cannot support the query function of the complex problem input by the user because the front and rear intentions of the complex problem input by the user cannot be associated. If the existing electronic map is used for inquiring the complex problems, the user is required to decompose the complex problems into a plurality of complex problems, and then the final inquiring result can be obtained through multiple inquiring operations. For example, an existing electronic map is used to query a complex problem of "how kender in the vicinity of a sea-letter building" and the implementation process is as follows: firstly, a sea message building is required to be searched on a hundred-degree map, then a user clicks to search nearby, and Kendeck is input, so that the inquiry function is finally realized.

In summary, how to improve the accuracy of identifying the user intention and reduce the man-machine interaction times in the query scene of the complex problem is still to be further researched.

Disclosure of Invention

The embodiment of the application provides a traffic travel problem query method and a traffic travel problem query device, which are used for improving the accuracy of identifying user intention and reducing man-machine interaction times.

In a first aspect, an embodiment of the present application provides a traffic travel problem query method, where the method includes: receiving traffic information to be queried; identifying a query intention corresponding to traffic information to be queried; extracting first descriptive information corresponding to the descriptive information type from traffic information to be queried according to the descriptive information type corresponding to the query intention; the first descriptive information includes location descriptive information and relationship descriptive information; searching a starting point entity corresponding to the position description information from the traffic knowledge graph; the traffic knowledge graph comprises entities of each layer in a traffic network and connection relations among the entities; determining a target entity according to the relation description information, the starting point entity and the connection relation among the entities included in the traffic knowledge graph; and deducing a target result corresponding to the query intention according to the traffic knowledge graph, the starting point entity and the target entity.

By the scheme, the query intention can be identified from the traffic information to be queried, and the first description information, such as the position description information and the relation description information, corresponding to the description information type can be extracted from the traffic information to be queried according to the description information type corresponding to the query intention. The traffic knowledge graph comprises entities of each layer in the traffic network and connection relations among the entities, so that the front and back intention of the associated user can be realized through the traffic knowledge graph, and the starting point entity corresponding to the position description information and the target entity are searched from the traffic knowledge graph. And then, according to the traffic knowledge graph, the starting point entity and the target entity, deducing a target result corresponding to the query intention. Therefore, the scheme provided by the application can identify the query intention of the user, realize the aim result corresponding to the query intention through the traffic knowledge graph starting point entity and the aim entity and deduce the aim result corresponding to the query intention, and does not need the user to decompose the traffic information to be queried like the prior art, thereby improving the accuracy of identifying the intention of the user and reducing the man-machine interaction times.

In a second aspect, an embodiment of the present application provides a traffic trip problem query device, including a memory and a processor; the memory is used for storing instructions; the processor is configured to execute the memory-stored instructions, which when executed by the processor, cause the apparatus to:

receiving traffic information to be queried; identifying the query intention corresponding to the traffic information to be queried; extracting first description information corresponding to the description information type from traffic information to be queried according to the description information type corresponding to the query intention; the first descriptive information comprises position descriptive information and relation descriptive information; searching a starting point entity corresponding to the position description information from a traffic knowledge graph; the traffic knowledge graph comprises entities of each layer in a traffic network and connection relations among the entities; determining a target entity according to the relation description information, the starting point entity and the connection relation among the entities included in the traffic knowledge graph; and deducing a target result corresponding to the query intention according to the traffic knowledge graph, the starting point entity and the target entity.

In a third aspect, an embodiment of the present application provides a traffic trip problem query device, including a receiving unit and a processing unit; the receiving unit is used for receiving traffic information to be queried;

the processing unit is used for identifying the query intention corresponding to the traffic information to be queried; extracting first description information corresponding to the description information type from traffic information to be queried according to the description information type corresponding to the query intention; the first descriptive information comprises position descriptive information and relation descriptive information; searching a starting point entity corresponding to the position description information from a traffic knowledge graph; the traffic knowledge graph comprises entities of each layer in a traffic network and connection relations among the entities; determining a target entity according to the relation description information, the starting point entity and the connection relation among the entities included in the traffic knowledge graph; and deducing a target result corresponding to the query intention according to the traffic knowledge graph, the starting point entity and the target entity.

In a fourth aspect, embodiments of the present application provide a computer storage medium having instructions stored therein which, when run on a computer, cause the computer to perform the method of the first aspect or any of the possible implementations of the first aspect.

In a fifth aspect, embodiments of the application provide a computer program product comprising instructions which, when run on a computer, cause the computer to perform the method of the first aspect or any of the possible implementations of the first aspect.

Drawings

FIG. 1 is a schematic diagram of a data structure suitable for use in an embodiment of the present application;

FIG. 2 is a schematic diagram of a traffic knowledge graph structure, to which the embodiment of the application is applicable;

FIG. 3 is a schematic diagram of a traffic search engine according to an embodiment of the present application;

fig. 4 is a flow chart of a traffic trip problem query method provided by an embodiment of the present application;

FIG. 5 is a schematic diagram of an entity relationship according to an embodiment of the present application;

FIG. 6 is a schematic diagram of another entity relationship provided in an embodiment of the present application;

FIG. 7 is a schematic diagram of another entity relationship provided in an embodiment of the present application;

fig. 8 is a schematic diagram of a route planning scenario provided in an embodiment of the present application;

fig. 9 is a schematic structural diagram of a traffic trip problem query device according to an embodiment of the present application;

fig. 10 is a schematic structural diagram of another route planning apparatus according to an embodiment of the present application.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings. The specific method of operation in the method embodiment may also be applied to the device embodiment or the system embodiment.

In order to solve the problem that the conventional electronic map cannot accurately identify the intention of a complex question method of a user, the embodiment of the application provides a traffic travel problem query method, which adopts a knowledge graph technology to realize knowledge expression of a traffic network, obtains a traffic knowledge graph, then identifies query intention, and combines the traffic knowledge graph to realize the effects of accurately identifying the intention of the user and reducing interaction times.

The knowledge graph is a large semantic network, and is used for describing the relationship between the entity concepts and the entity events of the objective world, specifically, the relationship is taken as a node, and the entity concepts are taken as an edge, so that a world mode is provided from the perspective of the relationship. Knowledge extraction is mainly open-ended link data, and typical inputs are natural language text or multimedia content documents. And then extracting available knowledge units by an automatic or semi-automatic technology, wherein the knowledge units mainly comprise 3 knowledge elements of entities, relations and attributes, and forming a series of high-quality fact expressions based on the knowledge elements.

In order to implement the method provided by the embodiment of the application, a traffic knowledge graph is required. Optionally, the traffic knowledge graph may be constructed by an electronic device when the traffic knowledge graph is needed, for example, the traffic knowledge graph is constructed according to a traffic network before the traffic knowledge graph is used, the traffic knowledge graph may also be configured in the electronic device in advance, whether the traffic knowledge graph is preset or the electronic device is reconstructed when the traffic knowledge graph is needed, and the construction process of the traffic knowledge graph may refer to the construction process of the traffic knowledge graph provided by the application.

The following describes the construction process of the traffic knowledge graph in detail, and specifically includes a first process and a second process.

In a first process, a data structure is used to describe a traffic network. Each type of information in the information of roads, intersections, geographical locations, regional divisions, etc. in the traffic network (map space) is abstracted to an entity describing the information, for example, a road entity "Jin Shuilu" entity, "215-way" bus entity, "sea information building" entity, etc.

Categorizing entities abstracted by the traffic network may include, but is not limited to, the following five types of entity information: traffic network, traffic nodes, points of interest, regional network, and dynamic information. As shown in fig. 1, the data contents included in the five types of entity information are as follows:

traffic network, such entity information describes route information of the traffic network, such as subway lines, bus lines, road networks, road sections, wherein entities included in the road network can describe minimum granularity of lanes, road directions, etc., and the whole road is an entity which cannot describe the information in detail.

The traffic nodes, the entity information of which describes the key entity information in the traffic network, can be the key nodes related to the traffic road network layer entities, such as subway stations, bus stations and road intersections.

The interest points, such entity information describes geographical location entities of interest to the user in the traffic network, such as dining, accommodation, shopping, health social security, scientific and educational culture, sports and leisure, tourist attractions, company enterprises, place names, etc., and are used in the traffic network to describe the starting point, ending point, access point, etc. of the user.

Regional networks, such entity information describes regional information in a traffic network, such as business circles, administrative areas, and the like.

Dynamic information, such entity information organizes information describing road congestion, dynamic events, dynamic road segments in the traffic network.

And step two, constructing a traffic knowledge graph according to the data structure of the step one. Specifically, an entity information traffic network is taken as a knowledge graph bottom layer, and a five-layer traffic knowledge graph structure is constructed by combining entity information such as interest points (Point of Interest, POIs), regional networks and dynamic information. Each layer of the traffic knowledge graph consists of a series of entities belonging to the layer classification, and a cross-layer connection mode is adopted for the connection mode between the entities, that is, two entities in the same layer are not connected in a specific connection relationship.

As shown in fig. 2, the traffic knowledge graph includes five layers, which are administrative district layer (may also be referred to as regional layer), POI layer, road intersection layer, road network layer-road section (may also be referred to as road section layer), road network layer-road (may also be referred to as road section layer), respectively. Any two entities on the same layer have no connection relationship, and the administrative district layer may have connection relationship with the entity of any one layer of the POI layer, the road intersection layer, the road network layer and the road section.

The administrative district layer may be a business district or administrative district. For example, the administrative district has business district a, business district B and administrative district C, roads and the like included in the business district a and the administrative district C may overlap, roads and the like included in the business district B and the administrative district C may overlap, but there is no connection relationship among the business district a, the business district B and the administrative district C on the administrative district layer, and only a connection relationship exists with a layer below the administrative district layer. Of course, two different entities belonging to the same layer may also be connected to the same entity of other layers at the same time, e.g. business district a and administrative district C may both be connected to the same POI entity in the POI layer.

The POI layer comprises a plurality of POI entities, and in the geographic information system, one POI entity can be a house, a shop, a mailbox, a bus station and the like.

The road intersection layer includes a plurality of entities, for example, the road intersection layer shown in fig. 2 includes 5 entities, namely, entity a, entity b, entity c, entity d, and entity e. Taking entity b as an example, entity b connects Shandong road 1, shandong road 2, yanji road 1, yanji road 2 in road network layer-road section.

The road network layer may be divided into a road network layer-road section and a road network layer-road, and one road may include a plurality of road sections, for example, a mountain road includes a mountain road 1 and a mountain road 2, as shown in fig. 2, a mountain road entity in the road network layer-road is connected to the mountain road 1 and the mountain road 2 in the road network layer-road, respectively, and a yan road entity in the road network layer-road is connected to the yan road 1 and the yan road 2 in the road network layer-road, respectively.

In the traffic network knowledge graph, the following connection modes can be adopted to realize knowledge graph construction:

in the first connection mode, the road network is connected with the road and the road section according to the attribution relationship, for example, the Shandong road 1 and the Shandong road 2 are connected with the Shandong road, and the connection mode of 'entity-relationship-entity' in the traffic knowledge graph is adopted. The Shandong roads 1 and 2 are road section identifiers above the Shandong roads, and the Shandong roads are specific roads included in the road network.

The second connection mode is that the road section can be connected with the road intersection, and the connection relation between the road section and the intersection in the road network is represented by adopting an entity-relation-entity in the connection mode according to the position of the road section in the traffic network.

And the third connection mode, wherein the interest point information is connected with road sections to represent the section of the road on which the interest point is in the road network, and the connection mode adopts an entity-relation-entity. The same interest point can be connected with a plurality of road sections, for example, the interest point near a road intersection, and the interest point is connected with two adjacent roads in an entity-relation-entity mode.

And in a fourth connection mode, the area network is connected with the interest point information, the road intersection and the road section, and the attribution information of the interest point and the road intersection is described, for example, a sea signal building belongs to the urban south area.

And the fifth connection mode is characterized in that the subway route and the bus route are respectively described as entities, and are connected with the described subway platform and bus platform by adopting an entity-relationship-entity connection mode.

And the sixth connection mode is that the subway platform and the bus platform are connected with the road section, and the connection mode of entity-relation-entity is adopted to represent the road on which the subway platform and the bus platform are located. For the same bus station related multi-path bus route, adopting multi-path public traffic and the bus station 'entity-relation-entity' connection mode to characterize; the same subway platform is positioned on different road sections (A, B, C, D ports), the platform is respectively connected with a plurality of road sections 'entity-relation-entity', the specific road sections where different exits are positioned are characterized by adopting the mode of 'entity-attribute value', and the attribute information of the bus platform entity is represented.

Specific information describing each entity, such as road length, subway platform longitude and latitude, bus line departure time, shopping square business hours and the like, is connected with the entity by adopting an entity-attribute value mode, and describes the detailed attribute information of the entity.

Based on the traffic knowledge graph, fig. 3 is a schematic structural diagram of a traffic travel search engine according to an embodiment of the present application.

As shown in fig. 3, the traffic travel search engine may include a web page (web) front end or application APP, a host interface (tkg _core) module, a voice recognition (voice_recovery) module, a multi-task semantic calculation (semanteme) module, a multi-task control (task_ctl) module, an alarm recognition (alarm_place) module, a real-time road condition inference search (real_time_traffic) module, a travel mode inference search (path_play) module, a geographic location inference search (query_information) module, a generalized semantic calculation (semantic) module, a data interface processing (data_drive_port) module, and other data interface modules. The main control interface module, the voice recognition module and the Toronesetting calculation module are mainly used for completing user question preprocessing and question service division, and the Toronesetting calculation module schedules a service processing algorithm according to the service division, and the generalization semantic calculation module and the data interface processing module complete other processing of service processing results. The functions of the respective service modules are described in detail below.

The main control interface module can be used for receiving a command triggered by the input of information to be queried by a web front end or a user in the electronic map application APP, and of course, the command can also be directly input by the user in the electronic map application APP; and if the data content carried by the command is text data, the main control interface module calls the Torons semantic calculation module to realize the identification of the user intention. On the other hand, the processing module may be further configured to receive a processing result, such as a route planning result, from the data interface processing module, and return the processing result to the web front end or the electronic map application APP.

And the voice recognition module is used for processing the voice data into text data.

The Toronchim semantic computation module is used for conducting Toronchim semantic computation, question-method storage, business division and the like, wherein the multiple rounds of semantic computation are used for identifying user intentions, such as candidate intentions, real-time road condition inquiry, travel mode inquiry, geographic position inquiry and the like, in any one or more of alarm identification, taking the four types of candidate intentions as examples, the Toronchim semantic computation is used for computing the probabilities that the user intentions are the four types of candidate intentions respectively according to the traffic information to be queried input by the user, and then sending the probabilities corresponding to the candidate intentions to the multi-task control module.

The multi-task control module is used for calling the service modules such as the alarm identification module, the real-time road condition reasoning and searching module, the trip mode reasoning and searching module, the geographic position reasoning and searching module and the like according to the probability result of semantic calculation. In an alternative mode, according to the order of probability from large to small, the candidate intention is taken as the user intention to call the corresponding service module, if any service module cannot infer that the result of meeting the condition corresponding to the service module is met, the next service module is called according to the probability order until one service module can obtain the result of meeting the condition corresponding to the service module, or none of the four service modules can obtain the result of meeting the condition corresponding to the corresponding service module. Specifically, the candidate intention with the highest probability is selected as the user intention, and then the traffic information to be queried is sent to the corresponding module for subsequent reasoning. If the candidate intention with the highest probability is taken as the user intention, the result cannot be inferred, one candidate intention with the highest probability is selected from the rest candidate intentions to be taken as the user intention, and then the traffic information to be queried is sent to the corresponding module for subsequent reasoning.

As an example, if it is determined that the user intends to query in a trip mode, the Toronesetting calculation module sends the traffic information to be queried to the trip mode reasoning search module. For another example, if it is determined that the user intends to query the real-time road condition, the Toronesetting calculation module sends the traffic information to be queried to the real-time road condition reasoning search module.

If all the service modules cannot infer the result, the generalized semantic computation module is called to determine a recommended question method and return the recommended question method to the main control interface module.

The following describes the service modules corresponding to candidate intentions such as alarm ground identification, real-time road condition inquiry, travel mode inquiry, geographic position inquiry and the like.

And the alarm identification module is used for identifying the alarm ground, and outputting the result that the user inputs the longitude and latitude of the geographic position coordinate point of the question method.

The real-time road condition reasoning searching module is used for carrying out reasoning searching of the real-time road condition, for example, by combining a traffic knowledge graph and a question method input by a user, reasoning the road section information to be queried by the user and requesting the real-time road condition interface to return the road congestion information of the road section to be queried.

And the travel mode reasoning search module is used for completing route planning of user intention under a certain travel mode of the user, such as a certain vehicle by combining with reasoning search of the knowledge graph. For example, "drive to kender's about a sea building," the vehicle of this question is a car, and finally a plurality of driving routes can be deduced, wherein shortest route, shortest route in use, high-speed priority route, etc. may be recommended.

And a geographic location reasoning search (query_information) module, which is used for searching the detailed information of the geographic location and outputting information such as the name, telephone, detailed address, type and the like of the user input question method.

If the processing result can be inferred through the corresponding business module after the determined user intention, the processing result enters the generalized semantic computation module from the corresponding inference module. If all the service modules cannot infer the result, the generalized semantic computation module is called to determine a recommended question method and return the recommended question method to the main control interface module.

A generalized semantic computation (generalized_semantic) module, configured to, when one service module exists among the several service modules and can correctly return a result, send the result to a data interface processing module; and if the service modules can not correctly return the results, returning a recommended question method to the data interface processing module, and transmitting the recommended question method to the main control interface module through the data interface processing module.

And the data interface processing (data_drive_port) module is used for carrying out standardized processing on the processing result, such as packaging the processing result into a fixed format, and returning the packaged content to the main control interface module so that the main control interface module outputs the processing result to the web front end or the electronic map application APP. In addition, for the additional requirement of other external systems on data, unified output processing can be performed through the data interface processing module.

The traffic travel question-answering engine based on the traffic network knowledge graph technology has the advantages that the man-machine interaction is more natural and more accurate, and compared with the traditional query system based on the map engine technology and the database technology, the traffic travel question-answering engine has the following advantages: on the one hand, the knowledge map expression of traffic roads, routes and geographic positions can realize the digitization and topology of map space data, and is convenient for data storage and retrieval. On the other hand, the traffic network expressed by the knowledge graph technology can provide queries such as fuzzy geographic positions, line transfer, places in the nearby range and the like based on topological relation for user travel questions and answers; on the other hand, the data maintenance is convenient, and the traffic network described by adopting the knowledge graph technology is used for maintaining the data such as newly-added buses, subway lines, newly-added roads and the like, namely, newly-added entities or newly-added attribute relation links; the route change is the attribute relation link change.

In the embodiment of the application, the traffic network knowledge graph can be integrated in the alarm ground identification module, the real-time road condition reasoning and searching module, the travel mode reasoning and searching module and the geographic position reasoning and searching module of the traffic travel search engine.

Based on the foregoing, fig. 4 illustrates a traffic issue query method according to an embodiment of the present application, where the method may be executed by the traffic search engine shown in fig. 3 and may also be executed by an electronic device including the traffic search engine. As shown in fig. 4, the method includes:

Step 401, receiving traffic information to be queried.

The traffic information to be queried can be a simple question method or a complex question method.

As described in connection with fig. 3, the traffic information to be queried may be entered by a user on the front end of the electronic device or on the electronic map application APP.

The traffic information to be queried can be voice data or text data, and if the traffic information to be queried is the text data, the main control interface module sends the received traffic information to be queried to the multi-round semantic computation module for processing so as to determine the query intention. If the traffic information to be queried is voice data, the main control interface module sends the received traffic information to be queried to the voice recognition module for processing so as to be processed into text data, and then sends the text data to the multi-round semantic computation module.

Step 402, identifying a query intention corresponding to traffic information to be queried.

Here, the query intent corresponding to the traffic information to be queried may include, but is not limited to, one or more of the following: alarm ground identification, real-time road condition inquiry, travel mode inquiry and geographic position inquiry.

For example, the traffic information to be queried can be classified by adopting a short text classification algorithm aiming at the traffic information to be queried, and the probability that the query intention is a certain type of intention is determined.

Step 403, extracting first description information corresponding to the description information type from the traffic information to be queried according to the description information type corresponding to the query intention, wherein the first description information comprises position description information and relation description information.

Alternatively, the description information type may include, but is not limited to, a location description type, an area description type, and a travel selection description type. For example, in step 403, the type of the description information corresponding to the query intention is a location description type, and the first description information may be location description information; for another example, in step 403, the description information type corresponding to the query intention is a region description type, and the first description information may be region blur information.

Description information types corresponding to query intents and first description information corresponding to the description information types are described below in connection with examples.

For example, the query is intended to identify an alarm or to query a geographic location, and the description types that can be extracted include a location description type and an area description type, wherein the location description information corresponding to the location description type can be a location description point of a building, a cell, a company, a road intersection, a parking lot, a village commission and the like, and the area fuzzy information corresponding to the area description type can be an area, a vicinity, 50 meters to the east, a number of downstairs, a doorway and the like.

For another example, the query is intended to be a travel mode query, and the description types which can be extracted include a location description type, an area description type and a travel mode description type, wherein the location description information corresponding to the location description type can be the point description information such as a starting point, an ending point, a passing point and the like of a building, a district, a company, a road intersection, a parking lot, a village commission and the like, and the area fuzzy information corresponding to the area description type can be the area fuzzy information such as an area, a nearby area, an eastern 50 m area, a number of downstairs, a gate and the like; riding, public transportation, subway, renting, self-driving, walking and other traffic modes; the travel selection description information corresponding to the travel mode description type can be the nearest travel, the nearest approach, the nearest tour, the nearest money saving, the nearest time saving, the nearest walking, and the like.

For another example, the query intention is real-time road condition query, and the description types which can be extracted include a road description type, a location description type and an area description type travel mode description type, wherein the road description information corresponding to the road description type can be a road, a road section and the like; the location description information corresponding to the location description type can be location description points of buildings, cells, companies, roads, road intersections, parking lots, village committee and the like; the region blur information corresponding to the region description type may be a region, a vicinity, a doorway, or the like.

In an alternative embodiment, an algorithm may be used to implement the extraction of the first description information corresponding to the description information type, such as a sentence pattern matching algorithm, a word segmentation decision algorithm, and a part-of-speech analysis algorithm. The sentence pattern matching algorithm can determine the question type of traffic information to be queried, such as a back question sentence pattern. The word segmentation decision algorithm is used for splitting traffic information to be queried into words. The part-of-speech analysis algorithm is used for synthesizing words and sentence patterns obtained by the first two algorithms and analyzing description information of which description type the words belong to.

Step 404, searching a starting point entity corresponding to the position description information from the traffic knowledge graph; the traffic knowledge graph comprises entities of each layer in the traffic network and connection relations among the entities.

As can be seen in conjunction with fig. 2, the traffic network included in the traffic knowledge graph may include five layers of entities, namely, a administrative district layer, a point of interest layer, a road intersection layer, a road layer, and a road section layer. Optionally, before the traffic knowledge graph is retrieved, a search index may be constructed between the location description information and the entity, so as to speed up searching for the corresponding entity according to the location description information.

And step 405, determining a target entity according to the relationship description information, the starting point entity and the connection relationship among the entities included in the traffic knowledge graph.

The determination of a target entity is described below in connection with specific examples.

Taking the relationship description information as a connection relationship as an example, the following problems can be handled: for example, the question methods of which stations the 26 road vehicles have, which stores the world wide, which transfer 214 road the 26 road vehicles are in are processed, are not limited to the above-mentioned question methods. As shown in fig. 5, the starting entity is entity 1, the target entity is entity 2, and the knowledge-graph reasoning is performed according to the connection relationship between the entity and other entities. Further, if the user problem includes that the entity 3 also needs to be inferred from the entity 2, the entity 3 may be inferred from the relationship 2.

Taking the example of which stores exist in the world as an example, the starting entity is the world, the target entity is the store, and the connection relationship between the starting entity and the target entity is the inclusion relationship.

Taking the relationship description information as the range information as an example, for example, the kender's question-about method processing near the sea-letter building, as shown in fig. 6, the starting point entity is taken as the entity 1, the target entity is taken as the entity 2, and the reasoning can be performed according to the geographical position proximity relationship between the entities. Taking as an example how kender's nearby a sea-mail building is going, the starting entity is the sea-mail building, the target entity is kender, and the range information is nearby.

Taking the relationship description information as a combination relationship, for example, the combination relationship of the connection relationship and the range information, for example, the question-law processing such as the Madao near the road junction of Nanjing road Ningxia. Firstly, a starting point entity is obtained by reasoning according to the connection relation, and then reasoning is carried out according to the range information according to the starting point entity and other entities. As shown in fig. 7, the entity connected with the entity 2 and the entity 3 simultaneously is adopted to locate the target entity as the entity 1, and the entity 4 is obtained by combining range reasoning. Taking the mcdonald effort near the road junction of south Beijing as an example, the south Beijing as an entity 1, the Ningxia way as an entity 2, the road junction of south Beijing as an entity 3, the mcdonald effort as an entity 4, and the range information as a neighborhood.

Of course, the combination relationship in the above example may be a combination of the connection relationship and the range information, or may be a combination of the connection relationship and the connection relationship, that is, the connection relationship between the entity 3 and the entity 4 in fig. 7 is also inferred. Alternatively, a combination of the connection relationship and the attribute information is also possible, that is, in fig. 7, the attribute relationship between the entity 4 and the entity 3, for example, the attribute information of the entity 4 is the attribute information of the entity 3.

And step 406, deducing a target result corresponding to the query intention according to the traffic knowledge graph, the starting point entity and the target entity.

By the scheme, the query intention can be identified from the traffic information to be queried, and the first description information, such as the position description information and the relation description information, corresponding to the description information type can be extracted from the traffic information to be queried according to the description information type corresponding to the query intention. The traffic knowledge graph comprises entities of each layer in the traffic network and connection relations among the entities, so that the front and back intention of the associated user can be realized through the traffic knowledge graph, and the starting point entity corresponding to the position description information and the target entity are searched from the traffic knowledge graph. And then, deducing a target result corresponding to the query intention according to the traffic knowledge graph, the starting point entity and the target entity. Therefore, the scheme provided by the application can identify the query intention of the user, realize the aim result corresponding to the query intention through the traffic knowledge graph starting point entity and the aim entity and deduce the aim result corresponding to the query intention, and does not need the user to decompose the traffic information to be queried like the prior art, thereby improving the accuracy of identifying the intention of the user and reducing the man-machine interaction times.

In an alternative way, the traffic knowledge graph may be constructed by: extracting various traffic information in a traffic network and association relations among various traffic information; and constructing a traffic knowledge graph by taking entities corresponding to various traffic information as nodes and the association relationship between any two types of traffic information as edges. By way of example, various types of traffic information may include, but are not limited to, the above-mentioned traffic road network, traffic nodes, points of interest, area networks, dynamic information, and other entity information, and traffic information used to construct a traffic knowledge graph may be any one or more of the above-mentioned various types of traffic information. The specific construction process refers to the related description of the two processes of constructing the traffic knowledge graph, and is not repeated here.

Based on the above embodiment, the query intention includes travel mode query, and the embodiment of the present application provides a mode capable of implementing the above step 406, determines a target travel mode corresponding to traffic information to be queried, determines a target traffic network according to the target travel mode, and infers a target route corresponding to the query intention according to a starting point entity, a target entity and the target traffic network.

In an alternative implementation, the target route may be determined by: firstly, determining a first adjacent entity which is nearest to a starting point entity in a target traffic network and a second adjacent entity which is nearest to the target entity in the target traffic network; and then, according to the target traffic network, planning a first route between the starting point entity and the first adjacent entity, a second route between the first adjacent entity and the second adjacent entity, and a third route between the second adjacent entity and the target entity, wherein the first route, the second route and the third route are combined to form the target route.

In one example, the information to be queried may include a travel mode, and the travel mode determines a traffic network adopted when the route is planned, for example, the travel mode is walking, riding, self-driving, renting, and the route is planned by adopting a road network; for example, the travel mode is public transportation, and a public transportation network is needed; and for example, the traveling mode is a subway, and a subway network is needed.

Illustratively, taking transfer scenes such as buses, subways and the like as examples, switching is needed to be based on road networks from a starting point to a public transportation network and from the public transportation network to an ending point. For example, a route planning from a place A to a place B takes a bus, and the middle needs to take 12 buses and then take 18 buses.

S1, carrying out range reasoning on a starting point, an ending point and adjacent target traffic mode network nodes by adopting a knowledge graph to obtain the nearest target traffic mode network entity. The method comprises the steps that a starting point is a place A, a destination point is a place B, a target travel mode is a bus, and first, 12 bus stops C near the place A and 18 bus stops F near the place B are found.

And S2, the route planning of the adjacent nodes of the starting point and the target traffic network and the route planning of the adjacent nodes of the ending point and the target traffic network are realized by adopting a road network from the starting point and the ending point. The road network is adopted to plan the route from the site A to the nearby 12-path bus station C and the route from the 18-path bus station F to the site B.

S3, performing head-to-tail route reasoning of the target traffic network on the target traffic network, wherein two situations are mainly adopted:

firstly, a common station transfer route planning is not needed between 12 buses and 18 buses, or the walking does not exceed a certain threshold value, and the threshold value can be determined according to actual needs, for example, the threshold value is set to be 20m, that is, the walking distance does not exceed 20m, and the common station transfer is realized; and is set to 5m, for example, without limitation. In this case, a bus network is used to plan a route between 12 bus stops C and 18 bus stops F according to the traffic knowledge graph.

And in the second case, a step transfer or a co-station transfer route planning with a step exceeding a certain threshold is needed between the 12 buses and the 18 buses. In this case, direct route reasoning cannot be completed, and route reasoning is performed by overlaying the road network. Assuming that 12 bus stops C to 12 bus stops D get off, and then 18 buses are transferred from 18 bus stops E to 18 bus stops F, according to a traffic knowledge graph, a route between the 12 bus stops C to 12 bus stops D is planned by adopting a bus network, a route between the 12 bus stops D to 18 bus stops E is planned by adopting a road network, and a route between the 18 bus stops E to 18 bus stops F is planned by adopting the bus network.

The route planning procedure is described below in connection with specific examples.

As shown in fig. 8, the following problems are addressed: what is the Zhejiang road Hubei road intersection walking to the happy guest to shop square? Planning is performed. Firstly, address reasoning is carried out, and a starting point and a terminal point are analyzed, wherein the starting point is a Zhejiang road Hubei road intersection, and the terminal point is a Happy guest shopping square. And then, carrying out position mapping in the traffic knowledge map, wherein the Zhejiang road, hubei road and crossing are mapped to a starting point entity a, and the Happy shopping square is mapped to a target entity d. And then carrying out route reasoning to obtain two routes of d-c-a and d-b-a.

Based on any embodiment, if no travel mode is provided in the traffic information to be queried, a route corresponding to a plurality of travel modes can be planned, so that a user can select one route by himself.

Further, after the route between the origin entity and the destination entity, a satisfactory route may be recommended according to the query intention of the user. For example, if the traffic information to be queried is how much money is needed from location point a to location point B, after the route between location point a and location point B is planned, multiple routes may be planned, and then a cheapest route may be recommended. If the traffic information to be queried is that the route from the position point A to the position point B is shortest, after the route between the position point A and the position point B is planned, a route with the shortest route can be recommended.

In a possible implementation manner, the traffic knowledge graph further comprises a correspondence between entities and attribute values. The embodiment of the application provides a mode capable of realizing the step 406, and the attribute value of the target entity corresponding to the query intention is deduced according to the corresponding relation between the entity and the attribute value, the starting point entity and the target entity included in the traffic knowledge graph. For example, a customer service call to a first department store is queried, the target entity is the first department store, and the attribute value is the telephone number of the customer service call.

Based on any of the foregoing embodiments, in one possible implementation manner, before identifying the query intent corresponding to the traffic information to be queried, a probability that the intent corresponding to the traffic information to be queried is a candidate intent may also be determined for each candidate intent of N candidate intents, where N is an integer greater than 0. And then determining the query intention according to the N probabilities corresponding to the determined traffic information to be queried.

Further, a plurality of ways of determining the query intention can be realized according to the determined N probabilities corresponding to the traffic information to be queried. In one possible implementation manner, the candidate intention corresponding to the maximum probability value in the N probabilities corresponding to the traffic information to be queried may be used as the query intention. And then proceeds to steps 403 to 406.

In yet another possible embodiment, the N candidate intents may be arranged in order of probability from large to small, and the i-th candidate intention is determined from the N candidate intents, where i is an integer smaller than N as the query intention. And then proceeds to steps 403 to 406.

Based on the second possible implementation manner, after deducing the target result corresponding to the query intention according to the traffic knowledge graph, the origin entity and the target entity, it may also be determined whether the target result meets the condition corresponding to the ith candidate intention.

Here, a case is exemplified in which the target result does not satisfy the condition corresponding to the i-th candidate intention. For example, the travel mode is a bus, but the output route is not a route using the bus as the travel mode; for another example, according to the information sent by the last service module, the operation is abnormal, and the currently input semantic information cannot be processed. For another example, a route abnormality such as kender from a cafe of a sea-going building to a sea-going building may not be satisfied in this case because the start point and the end point are together and the requirement is satisfied without planning.

Further, after determining whether the target route satisfies the condition corresponding to the ith candidate intention, the following two cases are specifically included:

if the target result is determined to not meet the condition corresponding to the ith candidate intention, selecting the (i+1) th candidate intention as the query intention;

and secondly, if the target result is determined to meet the condition corresponding to the ith candidate intention, outputting the target result.

Further, if any one of the N candidate intentions is used as the query intention, the determined target result does not meet the condition corresponding to the candidate intention, and prompt information is output; the prompt information is used for prompting the user to input recommended traffic information to be queried so as to determine a target entity according to the recommended traffic information to be queried.

Through the embodiment, the following effects can be achieved by combining the user intention recognition and travel question-answering engine design based on the traffic knowledge graph technology: for example, the complex operation problem of the user during traveling is solved, the user intention is recognized by a sentence, and the result is given; for another example, through multiple rounds of interaction, the user intent is precisely identified. When the user presents a problem and does not support a functional implementation, the user is prompted to enter missing information. For example, "how to walk from the fifth square to the new development center of the sea letter" prompts the user to input a transportation mode; as another example, generalized question-map reasoning. For example, "Kendeck near the south road and Ningxia road intersection" and deducing according to the knowledge graph to obtain the intersection which is the intersection of the south Beijing road and Ningxia road, and then identifying Kendeck near the intersection; for another example, the user inputs the association to identify the user intention, and inputs a question method of how to go for the place queried by the user, so that the route planning from the starting point to the end point of the user can be identified, and compared with map engines such as hundred degrees, the association of the user intention to the user intention can not be realized; for another example, the integration of information and data can be realized, and the user can be comprehensively covered for going out questions and answers.

Based on the same concept, fig. 9 illustrates a schematic structural diagram of a traffic problem query device according to an embodiment of the present application, and as shown in fig. 9, the device 900 may be used to execute any of the schemes shown in fig. 4. The apparatus 900 comprises a receiving unit 901 and a processing unit 902.

A receiving unit 901, configured to receive traffic information to be queried;

a processing unit 902, configured to identify a query intention corresponding to traffic information to be queried; extracting first descriptive information corresponding to the descriptive information type from traffic information to be queried according to the descriptive information type corresponding to the query intention; the first descriptive information includes location descriptive information and relationship descriptive information; searching a starting point entity corresponding to the position description information from the traffic knowledge graph; the traffic knowledge graph comprises entities of each layer in a traffic network and connection relations among the entities; determining a target entity according to the relation description information, the starting point entity and the connection relation among the entities included in the traffic knowledge graph; and deducing a target result corresponding to the query intention according to the traffic knowledge graph, the starting point entity and the target entity.

In an alternative embodiment, the query intent comprises a travel mode query; the processing unit is used for: determining a target travel mode corresponding to traffic information to be queried; determining a target traffic network according to a target travel mode; and deducing a target route corresponding to the query intention according to the starting point entity, the target entity and the target traffic network.

In an optional implementation manner, the traffic knowledge graph further comprises a corresponding relationship between the entity and the attribute value; the processing unit is used for: and deducing the attribute value of the target entity corresponding to the query intention according to the corresponding relation between the entity and the attribute value, the starting point entity and the target entity included in the traffic knowledge graph.

In an alternative embodiment, the processing unit is further configured to: determining the probability that the intention corresponding to the traffic information to be queried is the candidate intention aiming at each candidate intention in the N candidate intentions; n is an integer greater than 0; the processing unit is specifically configured to: and determining the query intention according to the N probabilities corresponding to the determined traffic information to be queried.

In an alternative embodiment, the processing unit is configured to: the N candidate intentions are arranged in the order from the big probability to the small probability, the ith candidate intention is determined from the N candidate intentions and used as the query intention, and i is an integer smaller than N; the processing unit is further configured to:

if the target result is determined to not meet the condition corresponding to the ith candidate intention, selecting the (i+1) th candidate intention as the query intention; and if the target result is determined to meet the condition corresponding to the ith candidate intention, outputting the target result.

In an alternative embodiment, the processing unit is specifically configured to: if any one of the N candidate intentions is used as the query intention, the determined target result does not meet the condition corresponding to the candidate intention, and prompt information is output; the prompt information is used for prompting the user to input recommended traffic information to be queried so as to determine a target entity according to the recommended traffic information to be queried.

In an alternative embodiment, the processing unit is specifically configured to: extracting various traffic information in a traffic network and association relations among the various traffic information; and constructing the traffic knowledge graph by taking the entity corresponding to the various traffic information as a node and the association relation between any two types of traffic information as an edge.

In an alternative embodiment, the various types of traffic information include any one or more of the following: traffic network, traffic nodes, points of interest, regional network, and dynamic information.

It should be understood that the above division of each unit is merely a division of a logic function, and may be fully or partially integrated into one physical entity or may be physically separated when actually implemented. In the embodiment of the present application, the receiving unit 901 and the processing unit 902 may be implemented by a processor 1002 in fig. 10 described below.

Based on the same concept, fig. 10 illustrates a schematic structural diagram of another traffic problem query device provided by the embodiment of the present application, and as shown in fig. 10, a device 1000 may be used to implement any of the schemes shown in fig. 4. The apparatus 1000 includes a memory 1001 and a processor 1002.

Memory 1001 may include volatile memory (RAM), such as random-access memory (RAM); the memory may also include a nonvolatile memory (non-volatile memory), such as a flash memory (flash memory), a hard disk (HDD) or a Solid State Drive (SSD); memory 1001 may also include a combination of the above types of memory.

The processor 1002 may be a central processing unit (central processing unit, CPU), a network processor (network processor, NP) or a combination of CPU and NP. The processor 1002 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (programmable logic device, PLD), or a combination thereof. The PLD may be a complex programmable logic device (complex programmable logic device, CPLD), a field-programmable gate array (field-programmable gate array, FPGA), general-purpose array logic (generic array logic, GAL), or any combination thereof. Alternatively, the memory 1001 and the processor 1002 may be integrated.

Alternatively, the memory 1001 may be used for storing program instructions, and the processor 1002 may invoke the instructions stored in the memory 1001, may perform one or more steps in the embodiment shown in the above-described scheme (e.g., the method shown in fig. 4), or an alternative implementation thereof, to enable the apparatus 1000 to implement the functions of the traffic problem querying apparatus in the above-described method.

A processor 1002 for executing the memory-stored instructions, which when executed by the processor 1002, cause the apparatus 1000 to: receiving traffic information to be queried; identifying the query intention corresponding to the traffic information to be queried; extracting first description information corresponding to the description information type from traffic information to be queried according to the description information type corresponding to the query intention; the first descriptive information comprises position descriptive information and relation descriptive information; searching a starting point entity corresponding to the position description information from a traffic knowledge graph; the traffic knowledge graph comprises entities of each layer in a traffic network and connection relations among the entities; determining a target entity according to the relation description information, the starting point entity and the connection relation among the entities included in the traffic knowledge graph; and deducing a target result corresponding to the query intention according to the traffic knowledge graph, the starting point entity and the target entity.

In an alternative embodiment, the query intent includes a travel mode query; the processor 1002 is specifically configured to cause the apparatus to: determining a target travel mode corresponding to the traffic information to be queried; determining a target traffic network according to the target travel mode; and deducing a target route corresponding to the query intention according to the starting point entity, the target entity and the target traffic network.

In an optional implementation manner, the traffic knowledge graph further includes a correspondence between entities and attribute values; the processor 1002 is specifically configured to cause the apparatus to: and deducing the attribute value of the target entity corresponding to the query intention according to the corresponding relation between the entity and the attribute value, the starting point entity and the target entity included in the traffic knowledge graph.

In an alternative embodiment, the processor 1002 is further configured to cause the apparatus to: determining the probability that the intention corresponding to the traffic information to be queried is the candidate intention aiming at each candidate intention in N candidate intentions; the N is an integer greater than 0; and determining the query intention according to the determined N probabilities corresponding to the traffic information to be queried.

In an alternative embodiment, processor 1002 is operative to cause an apparatus to perform the following operations: the N candidate intentions are arranged in the order from the big probability to the small probability, the ith candidate intention is determined from the N candidate intentions and used as the query intention, and i is an integer smaller than N; after a target result corresponding to the query intention is deduced according to the traffic knowledge graph, the starting point entity and the target entity, if the target result is determined not to meet the condition corresponding to the ith candidate intention, selecting the (i+1) th candidate intention as the query intention; and if the target result is determined to meet the condition corresponding to the ith candidate intention, determining the first route as the target result.

In an alternative embodiment, the processor 1002 is further configured to cause the apparatus to: if any one of the N candidate intentions is used as the query intention, the determined target result does not meet the condition corresponding to the candidate intention, and prompt information is output; the prompt information is used for prompting a user to input recommended traffic information to be queried so as to determine the target entity according to the recommended traffic information to be queried.

In an alternative embodiment, the processor 1002 is further configured to cause the apparatus to: extracting various traffic information in a traffic network and association relations among the various traffic information; and constructing the traffic knowledge graph by taking the entity corresponding to the various traffic information as a node and the association relation between any two types of traffic information as an edge.

In an alternative embodiment, the various types of traffic information include any one or more of the following: traffic network, traffic nodes, points of interest, regional network, and dynamic information.

In the above-described embodiments, may be implemented in whole or in part by software, hardware, firmware, or any combination thereof, and when implemented using a software program, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more instructions. When loaded and executed on a computer, produces a flow or function in accordance with embodiments of the present application, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The instructions may be stored in a computer storage medium or transmitted from one computer storage medium to another computer storage medium, for example, from one website, computer, server, or data center to another website, computer, server, or data center by a wired (e.g., coaxial cable, fiber optic, digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer storage media may be any available media that can be accessed by a computer or a data storage device such as a server, data center, or the like that contains an integration of one or more available media. The usable medium may be a magnetic medium (e.g., a floppy Disk, a hard Disk, a magnetic tape, a magneto-optical Disk (MO), etc.), an optical medium (e.g., CD, DVD, BD, HVD, etc.), or a semiconductor medium (e.g., ROM, EPROM, EEPROM, a nonvolatile memory (NAND FLASH), a Solid State Disk (SSD)), etc.

Embodiments of the present application are described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by instructions. These instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

It will be apparent to those skilled in the art that various modifications and variations can be made to the embodiments of the present application without departing from the spirit and scope of the application. Thus, if such modifications and variations of the embodiments of the present application fall within the scope of the claims and the equivalents thereof, the present application is also intended to include such modifications and variations.

Claims (10)

1. A traffic travel problem query method, comprising:

receiving traffic information to be queried;

calculating probabilities respectively corresponding to a plurality of candidate intentions according to the traffic information to be queried; sequentially determining candidate intentions serving as query intentions from large to small according to probabilities respectively corresponding to the candidate intentions; the candidate intents comprise one or more of alarm ground identification, real-time road condition inquiry, travel mode inquiry and geographic position inquiry;

identifying the query intent;

extracting first description information corresponding to the description information type from traffic information to be queried according to the description information type corresponding to the query intention; the description information type comprises at least one of a location description type, an area description type and a travel selection description type; the first descriptive information comprises position descriptive information and relation descriptive information;

Searching a starting point entity corresponding to the position description information from a traffic knowledge graph; the traffic knowledge graph comprises entities of each layer in a traffic network and connection relations among the entities; the traffic knowledge graph comprises a administrative district layer, a point of interest POI layer, a road intersection layer, a road network layer-road section, a road network layer-road; each layer of the traffic knowledge graph consists of a series of entities belonging to the layer classification, and a cross-layer connection mode is adopted for the connection mode between the entities;

determining a target entity according to the relation description information, the starting point entity and the connection relation among the entities included in the traffic knowledge graph;

and deducing a target result corresponding to the query intention according to the traffic knowledge graph, the starting point entity and the target entity.

2. The method of claim 1, wherein the query intent comprises a travel mode query;

the step of reasoning out the target result corresponding to the query intention according to the traffic knowledge graph, the starting point entity and the target entity, comprising:

determining a target travel mode corresponding to the traffic information to be queried;

determining a target traffic network according to the target travel mode;

And deducing a target route corresponding to the query intention according to the starting point entity, the target entity and the target traffic network.

3. The method of claim 1, wherein the traffic knowledge graph further comprises a correspondence of entities and attribute values;

the step of reasoning out the target result corresponding to the query intention according to the traffic knowledge graph, the starting point entity and the target entity, comprising:

and deducing the attribute value of the target entity corresponding to the query intention according to the corresponding relation between the entity and the attribute value, the starting point entity and the target entity included in the traffic knowledge graph.

4. The method of any one of claims 1-3, wherein prior to identifying the query intent corresponding to the traffic information to be queried, further comprising:

determining the probability that the intention corresponding to the traffic information to be queried is the candidate intention aiming at each candidate intention in N candidate intentions; the N is an integer greater than 0;

the identifying the query intention corresponding to the traffic information to be queried comprises the following steps:

and determining the query intention according to the determined N probabilities corresponding to the traffic information to be queried.

5. The method of claim 4, wherein the determining the query intent according to the determined N probabilities corresponding to the traffic information to be queried comprises:

the N candidate intentions are arranged in the order from the big probability to the small probability, and the ith candidate intention is determined from the N candidate intentions and used as the query intention, wherein i is an integer smaller than N;

the method further comprises the steps of, after reasoning out the target result corresponding to the query intention according to the traffic knowledge graph, the starting point entity and the target entity:

if the target result is determined to not meet the condition corresponding to the ith candidate intention, selecting the (i+1) th candidate intention as the query intention;

and if the target result is determined to meet the condition corresponding to the ith candidate intention, outputting the target result.

6. The method as recited in claim 5, further comprising:

if any one of the N candidate intentions is used as the query intention, the determined target result does not meet the condition corresponding to the candidate intention, and prompt information is output; the prompt information is used for prompting a user to input recommended traffic information to be queried so as to determine the target entity according to the recommended traffic information to be queried.

7. The method according to any one of claims 1-6, wherein the traffic knowledge graph is constructed by:

extracting various traffic information in a traffic network and association relations among the various traffic information;

and constructing the traffic knowledge graph by taking the entity corresponding to the various traffic information as a node and the association relation between any two types of traffic information as an edge.

8. The method of claim 7, wherein the types of traffic information include any one or more of:

traffic network, traffic nodes, points of interest, regional network, and dynamic information.

9. The traffic travel problem inquiry device is characterized by comprising a memory and a processor;

the memory is used for storing instructions;

the processor is configured to execute the memory-stored instructions, which when executed by the processor, cause the apparatus to perform the method of any of claims 1-8.

10. A computer storage medium having instructions stored therein which, when run on a computer, cause the computer to perform the method of any of claims 1-8.