CN111382252A

CN111382252A - Question category determination method, device, equipment and medium based on user request

Info

Publication number: CN111382252A
Application number: CN202010116522.3A
Authority: CN
Inventors: 林施扬; 邱岳亮; 王杰
Original assignee: Alipay Hangzhou Information Technology Co Ltd
Current assignee: Alipay Hangzhou Information Technology Co Ltd
Priority date: 2020-02-25
Filing date: 2020-02-25
Publication date: 2020-07-07
Anticipated expiration: 2040-02-25
Also published as: CN111382252B

Abstract

The embodiment of the specification discloses a problem category determining method based on a user request, which is used for acquiring the user request of a target user; if the historical request record of the target user exists, performing intention identification on the user request to obtain the intention of the target request of the user request; determining a target scene corresponding to the target request intention according to the historical request record of the target user and the target request intention; determining a target customer service robot corresponding to the target scene according to the corresponding relation between the scene and the customer service robot, and sending the user request to the target customer service robot; and if the historical request record of the target user does not exist, determining a target customer service robot corresponding to the user request from the M customer service robots according to at least one data of the target request intention and the embedded point information of the previous service page, and sending the user request to the target customer service robot.

Description

Question category determination method, device, equipment and medium based on user request

Technical Field

The embodiment of the specification relates to the technical field of customer service data processing, in particular to a problem category determination method, device, equipment and medium based on a user request.

Background

With the rapid development of the electronic information industry, more and more services are applied to electronic devices, and the more services, the more services have many associated services, and because the associated services have relevance, in order to reduce the number of applications in the electronic devices, a comprehensive service platform is usually used to provide customer service for the associated services.

The services provided by the existing integrated service platform are usually directly provided by related government departments or industry organizations in various cities, the business logic and the service conditions are different, and the integrated service platform needs to carry out sufficient business knowledge storage and time cost to build an intelligent customer service system covering all introduced services, and usually, the service platform is a self-owned customer service channel for manually guiding a client who complains or consults to an organization or an industry service provider.

Disclosure of Invention

The embodiment of the specification provides a problem category determination method, device, equipment and medium based on a user request, which can improve the matching degree of a target customer service robot and the user request, and further improve the accuracy of responding to the user request.

A first aspect of an embodiment of the present specification provides a method for determining a problem category based on a user request, including:

acquiring a user request of a target user;

if the historical request record of the target user exists, performing intention identification on the user request to obtain the intention of the target request of the user request; determining a target scene corresponding to the target request intention according to the historical request record of the target user and the target request intention; according to the corresponding relation between the scene and the customer service robots, determining a target customer service robot corresponding to the target scene from M preset customer service robots, and sending the user request to the target customer service robot, wherein M is an integer larger than 1.

If the historical request record of the target user does not exist, determining a target customer service robot corresponding to the user request from the M customer service robots according to at least one data of the target request intention and the embedded point information of the previous service page, and sending the user request to the target customer service robot, wherein the embedded point information of the previous service page is carried in the user request.

A second aspect of the embodiments of the present specification provides a method for determining a problem category based on a user request, which is applied to an integrated service platform, where the integrated service platform includes a platform customer service robot and can invoke a three-party customer service robot of a service provider hosted in the integrated service platform, and the content of a knowledge base relied on by the three-party customer service robot is maintained by the service provider corresponding to the three-party customer service robot, and the method includes:

acquiring a user request of a target user;

performing intention identification on the user request to obtain an intention identification result;

if the historical request record of the target user exists, determining a target scene corresponding to the user request according to the intention identification result and the historical request record of the target user;

if the historical request record of the target user does not exist and the intention identification result is that the intention of the target request is identified, acquiring a target scene corresponding to the user request according to the corresponding relation between the preset request intention and the scene; if the intention identification result is that the intention of the target request is not identified and the user request carries the embedded point information of the last service page, determining a target scene corresponding to the user request according to the embedded point information;

selecting the platform customer service robot or a certain three-party customer service robot as a target customer service robot according to the corresponding relation between a preset scene and the customer service robot; and sending the user request to the target customer service robot.

A third aspect of the embodiments of the present specification provides a method for determining a question category based on a user request, including:

acquiring a user request of a target user;

if the historical request record of the target user exists, determining a target scene corresponding to the user request according to the historical request record of the target user and the target request intention;

and determining a target customer service robot corresponding to the target scene from M preset customer service robots according to the corresponding relation between the preset scene and the customer service robots, and sending the user request to the target customer service robot, wherein M is an integer greater than 1.

A fourth aspect of the embodiments of the present specification provides an apparatus for determining a problem category based on a user request, including:

a user request obtaining unit, configured to obtain a user request of a target user;

the target customer service robot determining unit is used for identifying the intention of the user request to obtain the target request intention of the user request if the historical request record of the target user exists; determining a target scene corresponding to the target request intention according to the historical request record of the target user and the target request intention; determining a target customer service robot corresponding to the target scene from M preset customer service robots according to the corresponding relation between the scene and the customer service robots, wherein M is an integer greater than 1; if the historical request record of the target user does not exist, determining a target customer service robot corresponding to the user request from the M customer service robots according to at least one data of the target request intention and the embedded point information of the previous service page, wherein the embedded point information of the previous service page is carried in the user request

And the user request sending unit is used for sending the user request to the target customer service robot.

A fifth aspect of the embodiments of the present specification provides a device for determining a problem category based on a user request, which is applied to an integrated service platform, where the integrated service platform includes a platform customer service robot and is capable of invoking a three-party customer service robot that is hosted by a service provider of the integrated service platform, and the content of a knowledge base that the three-party customer service robot depends on is maintained by the corresponding service provider, and the device includes:

the request acquisition unit is used for acquiring a user request of a target user;

the intention identification unit is used for carrying out intention identification on the user request to obtain an intention identification result;

the target scene determining unit is used for determining a target scene corresponding to the user request according to the intention identification result and the history request record of the target user if the history request record of the target user exists; if the historical request record of the target user does not exist and the intention identification result is that the intention of the target request is identified, acquiring a target scene corresponding to the user request according to the corresponding relation between the preset request intention and the scene; if the intention identification result is that the intention of the target request is not identified and the user request carries the embedded point information of the last service page, determining a target scene corresponding to the user request according to the embedded point information;

the target robot determining unit is used for selecting the platform customer service robot or a certain three-party customer service robot as a target customer service robot according to the corresponding relation between a preset scene and the customer service robot;

and the request sending unit is used for sending the user request to the target customer service robot.

A sixth aspect of embodiments of the present specification provides an apparatus for determining a category of a problem based on a user request, including:

the data acquisition unit is used for acquiring a user request of a target user;

the intention determining unit is used for carrying out intention identification on the user request and obtaining an intention identification result;

the scene determining unit is used for determining a target scene corresponding to the user request according to the history request record of the target user and the target request intention if the history request record of the target user exists; if the historical request record of the target user does not exist and the intention identification result is that the intention of the target request is identified, acquiring a target scene corresponding to the user request according to the corresponding relation between the preset request intention and the scene; if the intention identification result is that the intention of the target request is not identified and the user request carries the embedded point information of the last service page, determining a target scene corresponding to the user request according to the embedded point information;

the robot determining unit is used for determining a target customer service robot corresponding to the target scene from M preset customer service robots according to the corresponding relation between the preset scene and the customer service robots, wherein M is an integer greater than 1;

and the sending unit is used for sending the user request to the target customer service robot.

The seventh aspect of the embodiments of the present specification further provides a system for determining a problem category based on a user request, including a client, a comprehensive service platform, and M customer service robots, where M is an integer greater than 1, including:

the client is used for acquiring input data of a target user, generating a user request and sending the user request to the comprehensive service platform;

the comprehensive service platform is used for receiving the user request sent by the client, and if a historical request record corresponding to the target user exists, performing intention identification on the user request to obtain a target request intention of the user request; determining a target scene corresponding to the target request intention according to the historical request record of the target user and the target request intention; (ii) a According to the corresponding relation between the scene and the customer service robots, determining a target customer service robot corresponding to the target scene from the M customer service robots, and sending the user request to the target customer service machine; if the embedded point information carried by the user request is obtained; determining a target customer service robot corresponding to the user request from the M customer service robots according to the embedded point information;

the target customer service robot is used for receiving the user request sent by the comprehensive service platform; responding the user request to obtain a response result; and returning the response result to the client through the integrated service platform.

The eighth aspect of the embodiments of the present specification further provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the problem category determination method based on the user request when executing the program.

The ninth aspect of the embodiments of the present specification also provides a computer-readable storage medium on which a computer program is stored, which, when executed by a processor, performs the steps of the above-described problem category determination method based on a user request.

The beneficial effects of the embodiment of the specification are as follows:

based on the technical scheme, after the user request of the target user is obtained, firstly, historical request record judgment is carried out, and when the historical request record exists, the target customer service robot is determined according to the historical request record and the target request intention identified by the intention, and the two constraint conditions, so that the accuracy of the determined target customer service robot is higher; even if the matching degree of the target customer service robot and the user request is higher, the accuracy of responding to the user request can be improved on the basis of higher matching degree of the target customer service robot and the user request.

In addition, when no history request record exists, a target customer service robot is determined according to at least one data of the target request intention and the embedded point information of the previous service page, the relevance between the previous service page and the user request is stronger because the previous service page is the previous service page requested by the user, and the relevance between the target scene and the user request is also enhanced by taking the service scene corresponding to the previous service page as the target scene on the basis of stronger relevance; and on the basis of stronger relevance between the target scene and the user request, the relevance between the determined target customer service robot and the user request is also strengthened, so that the accuracy of the determined target customer service robot is improved, namely, the matching degree between the target customer service robot and the user request is higher, and on the basis of higher matching degree between the target customer service robot and the user request, the accuracy of responding to the user request can also be improved.

Drawings

FIG. 1 is a flow chart of a method for determining a problem category based on a user request in an embodiment of the present disclosure;

FIG. 2 is a flowchart of a method for determining when a history request record of a target user exists in an embodiment of the present specification;

FIG. 3 is a flowchart illustrating a first method for determining when no historical request records exist for a target user in an embodiment of the present disclosure;

FIG. 4 is a flowchart illustrating a second method for determining when no historical request records exist for the target user in the embodiment of the present disclosure;

FIG. 5 is a schematic structural diagram of a problem category determination system based on a user request in an embodiment of the present specification;

FIG. 6 is a schematic diagram illustrating a connection structure between an integrated service platform and M customer service robots in an embodiment of the present disclosure;

FIG. 7 is a schematic structural diagram of data processing of an integrated service platform in an embodiment of the present disclosure;

FIG. 8 is a schematic structural diagram of an apparatus for determining a question category based on a user request in an embodiment of the present specification;

fig. 9 is a schematic structural diagram of an electronic device in an embodiment of this specification.

Detailed Description

In order to better understand the technical solutions, the technical solutions of the embodiments of the present specification are described in detail below with reference to the drawings and specific embodiments, and it should be understood that the specific features of the embodiments and embodiments of the present specification are detailed descriptions of the technical solutions of the embodiments of the present specification, and are not limitations of the technical solutions of the present specification, and the technical features of the embodiments and embodiments of the present specification may be combined with each other without conflict.

In a first aspect, as shown in fig. 1, an embodiment of the present specification provides a method for determining a problem category based on a user request, including:

s102, acquiring a user request of a target user;

s104, if the history request record of the target user exists, acquiring the history request record of the target user; performing intention identification on the user request to obtain a target request intention of the user request; determining a target scene corresponding to the target request intention according to the historical routing data and scene configuration; according to the corresponding relation between the scene and the customer service robots, determining a target customer service robot corresponding to the target scene from M preset customer service robots, and sending the user request to the target customer service robot, wherein M is an integer larger than 1.

S106, if the historical request record of the target user does not exist, determining a target customer service robot corresponding to the user request from the M customer service robots according to at least one data of the target request intention and the embedded point information of the previous service page, wherein the embedded point information of the previous service page is carried in the user request.

The problem category determination method based on the user request provided by the embodiment of the specification is applied to a comprehensive service platform, each customer service robot corresponds to at least one service, namely, one customer service robot can process one or more services; further, the M customer service robots include a platform customer service robot of the integrated service platform and a third party customer service robot provided by a third party service provider.

In the embodiment of the present specification, the integrated service platform may be a tablet computer, a notebook computer, a desktop computer, an all-in-one machine, a smart phone, and other devices; further, the client may be a tablet computer, a notebook computer, a desktop computer, an all-in-one machine, a smart phone, and the like.

Specifically, when the third-party service party includes a public deposit service party, a social security service party and a gas service party, the M customer service robots include a public deposit service robot provided by the public deposit service party, a social security service robot provided by the social security service party, a gas service robot provided by the gas service party and the platform service robot.

In the embodiment of the specification, the integrated service platform provides a unified customer service interface for unified intelligent customer service; enabling a user to enter the unified customer service interface through a third-party specific service page and a platform page of the comprehensive service platform, receiving input data of the user on the unified customer service interface by a client, and generating the user request according to the input data; and the client sends the generated user request to the comprehensive service platform, so that the comprehensive service platform acquires the user request.

In this embodiment of the present specification, the history request record of the target user and the user request of the target user may be requests generated in one session, and at this time, the target user sends multiple requests including the user request and the history request in one session; of course, the history request record of the target user and the user request of the target user may be requests generated by the target user in multiple sessions.

The first session in the embodiment of the present specification is specifically a session when a user enters a unified customer service interface provided by the integrated service platform until the unified customer service interface exits, for example, the user enters the unified customer service interface at a first time and continues to exit at a second time, and at this time, the interaction between the user and the unified customer service interface from the first time to the second time is taken as the first session; and the user enters the unified customer service interface at a third time after the second time and exits at a fourth time, and at the moment, the interaction between the user and the unified customer service interface from the third time to the fourth time is used as another session. Therefore, each user request is generated by the user and the integrated service platform in the session process.

Thus, in step S102, the user request sent by the client may be received first; the user request may include not only the input data but also a target user identifier of the target user, where the user identifier may be user I D, and the user request may correspond to any one of the N services.

In an embodiment of this specification, after the client establishes a connection with the unified customer service interface, the client establishes a session channel with the integrated service platform, and the integrated service platform receives the user request sent by the client through the session channel.

Specifically, the user request further carries embedded point information of a previous service page, where the previous service page is a previous service page of the target user accessing the unified client-server interface, and the embedded point information includes service subject information of the previous server page, and the service subject information includes a subject of the previous server page. For example, social security queries provided in the pay for treasure city service block are provided by social security offices, and when a user is in doubt about the query content, the object to be consulted is a local social security office providing the social security query, and the social security office is the service subject.

In the embodiment of the present specification, the integrated service platform performs embedding on a scene where a user is located before entering the unified customer service interface, that is, embeds a previous service page of each user before accessing the unified customer service interface, and is configured to collect embedding point information of the previous service page and write the embedding point information into a user request of the user; in this way, the previous service page accessed before the target user accesses the unified customer service interface is the previous service page, and point burying is performed in the process that the target user accesses the previous service page; according to the embedding point in the process that the target user accesses the previous service page, the embedding point information of the target user accessing the previous service page can be obtained; thus, in the process of generating the user request according to the input data, the user request can be generated according to the input data and the buried point information; after the user request is generated, the embedded point information of the previous service page may also be directly written into the user request, and this specification is not limited in particular.

For example, a b-city public accumulation fund service page displayed on a client by a target user a enters the unified customer service interface, then input data of a in the unified customer service interface is acquired as 'the public accumulation fund account password is forgotten', a user request of a is generated according to the 'the public accumulation fund account password is forgotten', wherein the user request of a carries the data of 'the public accumulation fund account password is forgotten', the user I D of a is 10258 for example, and carries the buried point information of a accessing the b-city public accumulation fund service page; and then the client sends the user request of a to the comprehensive service platform, so that the comprehensive service platform acquires the user request of a.

After step S102 is executed, it may also be first detected whether a history request record of the target user exists in a platform database of the integrated service platform, and if the history request record of the target user exists in the platform database, step S104 is executed; if the historical request record of the target user does not exist in the platform database, step S106 is executed.

In an embodiment of the present specification, the platform database may be an electronic device such as a desktop computer, a notebook computer, a smart phone, an all-in-one machine, and the like, a history request record of a history user is stored in the platform database, the history user is a user before the target user accesses the integrated service platform, and the history request record records user request data, a subsequent processing result, and response content of the customer service robot; at this time, the history request record of each history user can be associated with the user identifier of the history user, so that the multiplexing of the user requests of the user in the subsequent round is facilitated.

Specifically, after the user request is acquired in step S102, according to the target user identifier carried in the user request, the target user identifier may be used as search content to search whether a history request record corresponding to the target user identifier exists in the platform database; at this time, it may be queried whether a history request record corresponding to the target user identifier exists in a session record in which the user request is located, or it may be queried whether a history request record corresponding to the target user identifier is inserted in all session records stored in the platform database.

At this time, when querying whether a history request record corresponding to the target user identifier exists in a session record in which the user request is located, if a history request record exists, determining that the history request record of the target user exists in the platform database, and then executing step S104; if the history request record does not exist, determining that the history request record of the target user does not exist in the platform database, and then executing step S106.

Further, when a history request record corresponding to the target user identifier is inquired in all session records stored in the platform database in an inserting manner, if the history request record exists in the inquiry, it is determined that the history request record of the target user exists in the platform database, and then step S104 is executed; if the history request record does not exist, determining that the history request record of the target user does not exist in the platform database, and then executing step S106.

If it is detected that the history request record of the target user exists in the platform database, step S104 includes the following steps, referring to fig. 2, which are specifically as follows:

step S1042, obtaining the history request record of the target user;

specifically, the historical request record of the target user can be searched from the platform database according to the target user identifier; if the history request record in the integrated service platform exists in an external database, the history request record of the target user can be searched from the external database.

Step S1044, identifying the intention of the user request to obtain the target request intention of the user request;

specifically, the target request intention may be obtained by performing intention recognition on the user request using a trained Natural Language Understanding (NLU) model.

Specifically, the trained NLU model may be obtained by training a plurality of history request records, and as the number of users accessing the integrated service platform increases, the trained NLU model may continue to be iterated and optimized by using the history request records corresponding to the increased users, so that the trained NLU model is optimized along with the increase of the user access amount, and the result identified by the optimized trained NLU model is more accurate.

And S1046, determining a target scene corresponding to the target request intention according to the historical request record of the target user and the target request intention.

Specifically, the correspondence between the history request intention of the target user and the history session scene may be acquired from the history request record of the target user, and first, whether there is an intention identical to the target request intention in the history request intention is detected, and if there is an intention identical to the target request intention, the target scene may be acquired from the correspondence between the history request intention and the history session scene; if not, acquiring the target scene according to a preset scene corresponding table and the target request intention; therefore, the target scene can be identified more quickly and accurately through the corresponding relation between the historical request intention and the historical conversation scene in the historical request record, and the accuracy of the identified target customer service robot can be improved on the basis of improvement of the identified target scene.

Specifically, a scene corresponding to the target request intention may be searched from the scene correspondence table as the target scene with the target request intention as a search condition.

Step S1048, according to the preset corresponding relation between the scene and the customer service robot, determining the target customer service robot corresponding to the target scene, and sending the user request to the target customer service robot.

In the embodiment of the present specification, the platform database may be preset with the scene correspondence table, and the correspondence between the scene and the customer service robot may be stored in the platform database through the scene correspondence table; in this way, the target customer service robot corresponding to the target scene can be obtained according to the scene corresponding table and the scene service corresponding table; then, the routing information of the target customer service robot is obtained, and the user request is sent to the target customer service robot through the routing information of the target customer service robot; and after the target customer service robot acquires the user request, responding to the user request to obtain a response result, and returning the response result to the client corresponding to the user request.

Specifically, the scene correspondence table has an account abnormal intention corresponding to a platform service scene, a b1 municipal public deposit account query intention corresponding to a b1 municipal public deposit service scene, and a b2 municipal social security account password abnormal intention corresponding to a b2 municipal social security service scene; correspondingly, the service correspondence table comprises a platform service scene corresponding to the platform customer service robot, a b1 market public deposit service scene corresponding to the b1 market public deposit customer service robot, and a b2 market social security service scene corresponding to the b2 market social security customer service robot.

For example, after receiving a user request of a sent by a client, the integrated service platform first queries in a platform database by using a user ID 10258 carried in the user request of a, and at this time, it may query whether a history request record with an ID of 10258 exists in a session record in which the user request of a is located, or insert a query whether a history request record with an ID of 10258 exists in all session records stored in the platform database; if the historical request record with the ID of 10258 is inquired, the historical request record with the ID of 10258 is obtained, then the intention recognition is carried out on the user request (the account password of the public fund is forgotten) of a through a trained NLU model, and the target request intention of a is the abnormal intention of the account password of the public fund in the city b.

Further, according to the history request record with the ID of 10258, determining the corresponding relation between the history request intention with the ID of 10258 and the history session scene, and if the corresponding relation between the abnormal intention of the b-city public deposit account password and the b-city public deposit service exists, acquiring a target scene corresponding to the user request with the ID of 10258 from the corresponding relation, namely the b-city public deposit service; and then searching a target customer service robot corresponding to the b-city public deposit service from the scene service corresponding table by taking the b-city public deposit service as a searching condition according to the scene service corresponding table stored in the platform database.

Of course, after determining the correspondence between the history request intention with ID 10258 and the history session scene according to the history request record with ID 10258, if the target scene corresponding to the user request with ID 10258 is not obtained from the correspondence, the target scene of a may be obtained according to the scene correspondence table in the platform database and the target request intention of a; and then according to the target scene of a and the scene service corresponding table in the platform database, acquiring that the target customer service robot of a is a B market accumulation fund customer service robot.

Further, after the target customer service robot is the b city public deposit customer service robot, the user request of the a is sent to the b city public deposit customer service robot according to the routing information of the b city public deposit customer service robot, the b city public deposit customer service robot responds to the user request of the a, and the response result is returned to the client side where the user request of the a is generated.

In addition, if the input data carried by the user request of the a is 'how the account can not be logged in' so as to hit the account abnormal intention, then the scene corresponding to the account abnormal intention is searched from the intention corresponding table to be a platform service scene, and the customer service robot corresponding to the platform service scene is searched from the image service corresponding table to be the platform customer service robot, at the moment, although the user a enters the unified customer service interface of the comprehensive service platform from the b city public fund page, the user request of the a is sent to the platform customer service robot instead of the b city public fund customer service robot; and then the platform customer service robot responds to the user request of the a, and returns a response result to the client side where the user request of the a is generated.

In this way, after the user request of the target user is obtained, firstly, historical request record judgment is carried out, and when the historical request record exists, the target customer service robot is determined according to the historical request record and the target request intention identified by the intention, and the two constraint conditions, so that the accuracy of the determined target customer service robot is higher; and when no history request record exists, determining the target customer service robot according to the embedded point information of the previous service page carried in the user request, so that the determined target customer service robot is determined according to the embedded point information of the previous service page, and the accuracy of the determined target customer service robot is higher.

If the history request record corresponding to the target user identifier is not searched from the platform database, it is determined that the history request record of the target user does not exist in the platform database, and then step S106 is executed, where in the process of executing step S106, since the history request record of the target user does not exist in the platform database, it may be determined that the target user accesses the integrated platform server for the first time, at this time, the following steps may be included, as shown in fig. 3:

and step S302, performing intention identification on the user request.

Specifically, the trained NLU model may be used to perform intent recognition on the user request, and determine whether the recognized intent is any one of a plurality of preset intentions; if the identified intention is any one of the preset intentions, taking the identified intention as the target request intention, and then sequentially executing the steps S304-S306; if the identified intention is not any one of the preset intentions, determining that the target request intention requested by the user is not identified, and then executing the steps S404-S406 in sequence.

For example, the preset intentions include an account abnormal intention, a public deposit account inquiry intention, a social security account password abnormal intention and a gas account inquiry intention, if the user request of the user a1 is identified through the NLU model, the identified intention is a tax inquiry intention, and the steps S404-S406 are executed because the tax inquiry intention is not in the preset intentions; if the user request of the user a1 is identified through the NLU model, the identified intent is an account abnormality intent, and the steps S304-S306 are further executed because the account abnormality intent is among the preset intentions.

Step S304, if a target request intention corresponding to the user request is identified, acquiring the target scene according to the scene correspondence table and the target request intention.

Specifically, if the target request intention is identified in step S302, the scene correspondence table is obtained from the platform database, and a query is performed in the scene correspondence table using the target request intention, so that a scene corresponding to the target request intention is found from the scene correspondence table as the target scene.

For example, the scene correspondence table has an account abnormal intention corresponding to a platform service scene, a b1 city public deposit account query intention corresponding to a b1 city public deposit service scene, and a b2 city social security account password abnormal intention corresponding to a b2 city social security service scene; if the user request of a1 is identified in step S302, and the identified target request intention is an account abnormal intention, the account abnormal intention is used as a search condition, and a corresponding scene is searched from the scene correspondence table as a platform service scene, so that the target scene can be determined as the platform service scene.

Step S306, according to the corresponding relation between the scene and the customer service robot, determining a target customer service robot corresponding to the target scene, and sending the user request to the target customer service robot.

In the embodiment of the present specification, the platform database may be preset with the scene correspondence table, and the correspondence between the scene and the customer service robot may be stored in the platform database through the scene correspondence table; in this way, the target scene can be used for querying in the scene service correspondence table, and the customer service robot corresponding to the target scene is searched from the scene service correspondence table to serve as the target customer service robot; then, the routing information of the target customer service robot is obtained, and the user request is sent to the target customer service robot through the routing information of the target customer service robot; and after the target customer service robot acquires the user request, responding to the user request to obtain a response result, and returning the response result to the client corresponding to the user request.

Specifically, the detailed implementation process of step S306 may refer to the description of step S1048, and for brevity of the description, the detailed description is omitted here.

For example, also taking the user request of the user a as an example, if the user ID carried in the user request of a is 10258, querying whether a history request record with ID 10258 exists; and if the historical request record with the ID of 10258 does not exist, performing intention identification on the user request of a by using an NLU model, wherein the identified intention is an account abnormal intention.

Further, if the identified account number abnormal intention is in the plurality of preset intentions, finding a scene corresponding to the account number abnormal intention from the scene corresponding table, wherein the scene corresponding to the account number abnormal intention is found to be a platform service scene because the account number abnormal intention is corresponding to the platform service scene in the scene corresponding table; and searching a service robot corresponding to the abnormal intention of the account number from the scene service corresponding table as a platform service robot, namely, a target service robot of the user a as the platform service robot, further sending the user request of the user a to the platform service robot, responding the user request of the user a by the platform service robot, and returning a response result to the client side where the user a is located.

In this way, when no historical request record exists, the intention recognition is performed on the user request, and the trained NLU model is utilized and trained on a plurality of historical request records, so that the target request intention recognized by the trained NLU model is more accurate, and the accuracy of the target customer service robot determined according to the target request intention is improved on the basis of more accurate target request intention.

And, if no target request intention corresponding to the user request is identified, including the following steps, as shown in fig. 4 in detail:

step S402, analyzing the user request to obtain the embedded point information of the previous service page, obtaining the service scene corresponding to the previous service page, and taking the service scene as the target scene.

In an embodiment of the present specification, the integrated service platform buries a point of a scene where a user is located before entering the unified customer service interface, and stores a point-scene mapping relationship between the buried point and the service scene in the platform database.

Specifically, the user request may be analyzed to obtain the embedded point information of the previous service page, the service scene may be obtained according to the embedded point location of the embedded point information and the point scene mapping relationship, and the service scene may be used as the target scene.

Specifically, after the embedded point information of the previous service page is obtained through analysis, the point scene mapping relationship is obtained from the platform database, then the embedded point where the embedded point information is located is used for querying in the point scene mapping relationship, and a queried scene is used as the service scene of the previous service page, that is, the queried scene is used as the target scene.

For example, the point-view mapping relationship includes that a scene corresponding to a b1 city public deposit service page with a buried point location of b1 is a b1 city public deposit service scene, a scene corresponding to a b2 city social security service page is a b2 city social security service scene, and a scene corresponding to a b3 city gas service page is a b3 city gas service scene; if the user request of the user a2 is analyzed, and the embedded point position where the embedded point information of the previous service page is located is the b3 city gas service page, the b3 city gas service page is used for inquiring in the point scene mapping relation, and the b3 city gas service scene is inquired to be used as the target scene of a 3.

Step S404, according to the corresponding relation between the scene and the customer service robots, obtaining the customer service robots corresponding to the service scene from the M customer service robots, using the obtained customer service robots as the target customer service robots, and sending the user request to the target customer service robots.

Specifically, the service robot corresponding to the service scene is acquired, the service robot corresponding to the service scene is set as the target service robot, and the user request is transmitted to the target service robot.

Specifically, after the customer service robot corresponding to the service scenario is taken as the target customer service robot, the routing information of the target customer service robot may be acquired, and the user request is sent to the target customer service robot through the routing information of the target customer service robot; and after the target customer service robot acquires the user request, responding to the user request to obtain a response result, and returning the response result to the client corresponding to the user request through the comprehensive service platform.

Specifically, the detailed implementation process of step S404 may refer to the description of step S1048, and for brevity of the description, the detailed description is omitted here.

In the embodiment of the present specification, when a certain customer service robot responds to the user request, first, according to the user request, searching whether a response result corresponding to the user request exists in a database of the customer service robot, and if so, directly returning the response result to a corresponding client through the integrated service platform; and if the response operation does not exist, determining one response operation from a plurality of preset response operations as a final response operation, responding to the user request by using the final response operation, obtaining the response result and returning the response result to the corresponding client.

In this embodiment of the present specification, after the target scene corresponding to the target request intention is obtained through steps S302 to S304 and the service scene of the previous service page is obtained through step S402, one scene is selected from the two scenes through the set selection condition as the target scene requested by the user, then the target customer service robot is determined according to the target scene requested by the user, and then the user request is sent to the target customer service robot.

In an embodiment of the present specification, the multiple response operations include manual response operations and no-answer response operations, and when the final response operation is a manual response operation, an artificial channel supported by a service provider corresponding to the target customer service robot is acquired, and the acquired artificial channel is packaged into a service card and rendered at a client, so that the target user enters a subsequent response step of the artificial channel according to the service card; when the final response operation is no-answer response operation, acquiring an artificial channel supported by a service provider corresponding to the target customer service robot, packaging the acquired artificial channel into a service card, and rendering the service card at a client so that the target user enters a subsequent response step of the artificial channel according to the service card; and the preset no-answer prompt information can be directly returned to the corresponding client, a no-answer work order is created, and then data analysis and service optimization are carried out on the no-answer work order.

Specifically, the response mode of the manual response operation is a manual bottom-entering mode, namely, an acquired manual channel is packaged into a service card to be rendered at a client, so that the target user enters a subsequent response step of the manual channel according to the service card; the answer-free response operation response mode can be a manual bottom-entering mode or a work order bottom-entering mode, wherein the work order bottom-entering mode is that the answer-free prompt information is directly returned, and then the answer-free work order is created to carry out data analysis and service optimization on operators in corresponding scenes.

In another embodiment of the present disclosure, a method for determining a problem category based on a user request is provided, which is applied to an integrated service platform, where the integrated service platform includes a platform customer service robot and can invoke a three-party customer service robot of a service provider hosted in the integrated service platform, and the content of a knowledge base relied on by the three-party customer service robot is maintained by the corresponding service provider, and the method includes the following steps:

step A1, obtaining a user request of a target user;

the specific implementation process of step a1 refers to the description of S102, and for brevity of the description, is not repeated herein.

Step A2, performing intention identification on the user request to obtain an intention identification result;

step A3, if the historical request record of the target user exists, determining a target scene corresponding to the user request according to the intention identification result and the historical request record of the target user;

the implementation of steps a2 and A3 refers to the description of step S104.

Step A4, if the historical request record of the target user does not exist and the intention identification result is that the target request intention is identified, obtaining a target scene corresponding to the user request according to the corresponding relation between the preset request intention and the scene; if the intention identification result is that the intention of the target request is not identified and the user request carries the embedded point information of the last service page, determining a target scene corresponding to the user request according to the embedded point information;

step A5, selecting the platform service robot or a certain three-party service robot as a target service robot according to the corresponding relation between a preset scene and the service robot; and sending the user request to the target customer service robot.

The implementation of steps a4 and a5 refers to the description of step S106.

Another embodiment of the present specification provides a method for determining a question category based on a user request, including:

step B1, acquiring a user request of a target user;

the specific implementation process of step B1 refers to the description of step S102, and for brevity of the description, is not repeated herein.

Step B2, performing intention identification on the user request to obtain an intention identification result;

step B3, if the historical request record of the target user exists, determining a target scene corresponding to the user request according to the historical request record of the target user and the target request intention;

the implementation of steps B2 and B3 refers to the description of step S104.

Step B4, if the historical request record of the target user does not exist and the intention identification result is that the target request intention is identified, obtaining a target scene corresponding to the user request according to the corresponding relation between the preset request intention and the scene; if the intention identification result is that the intention of the target request is not identified and the user request carries the embedded point information of the last service page, determining a target scene corresponding to the user request according to the embedded point information;

and step B5, according to the corresponding relation between the preset scene and the customer service robots, determining target customer service robots corresponding to the target scene from M preset customer service robots, and sending the user request to the target customer service robots, wherein M is an integer greater than 1.

The implementation of steps B4 and B5 refers to the description of step S106.

For example, also taking the user request of the user a as an example, if the user ID carried in the user request of a is 10258, querying whether a history request record with ID 10258 exists; and if the historical request record with the ID of 10258 does not exist, performing intention identification on the user request of a by using an NLU model, wherein the identified intention is a false intention of the accumulation fund account.

Further, if the identified false intention of the public accumulation fund account number is in the preset intentions, acquiring point burying information of a previous service page carried by the user request of the point a, and determining that a target scene of the point a is a b1 market public accumulation fund service scene because the point scene mapping relationship comprises that a scene corresponding to the public accumulation fund service page with a point burying position of b1 is a b1 market public accumulation fund service scene according to the point burying position of the point burying information of the previous service page being a b1 market public accumulation fund service page; and then searching a customer service robot corresponding to the b1 market public deposit service scene as the b1 market public deposit service robot from the scene service corresponding table, namely, the target customer service robot of the user a is the b1 market public deposit service robot, further sending the user request of a to the b1 market public deposit service robot, responding to the user request of a by the b1 market public deposit service robot, and returning the response result to the client where a is located.

And in the process that the b1 municipal backlog service robot responds to the user request of a, if a response result corresponding to the user request of a exists in the database of the b1 municipal backlog service robot, if the response result exists, the response result is directly returned to the corresponding client; if the response operation does not exist, acquiring a manual response operation selected by the user aiming at the multiple response operations as a final response operation, then acquiring a manual channel of a b1 market public deposit manager corresponding to the b1 market public deposit customer service robot, and packaging the acquired manual channel into a service card to render at the client, so that the user a enters a subsequent response step of the manual channel according to the service card.

Thus, when no history request record exists and the user request is subjected to intention identification to hit a preset intention, a service scene corresponding to the previous service page is acquired according to the embedded point information of the previous service page carried by the user request, the service scene is used as the target scene, the relevance between the previous service page and the user request is stronger because the previous service page is the previous service page of the user request, and the relevance between the target scene and the user request is also enhanced on the basis of stronger relevance by using the service scene corresponding to the previous service page as the target scene; and on the basis of stronger relevance between the target scene and the user request, the relevance between the determined target customer service robot and the user request is also strengthened, so that the accuracy of the determined target customer service robot is improved, namely, the matching degree between the target customer service robot and the user request is higher, and on the basis of higher matching degree between the target customer service robot and the user request, the accuracy of responding to the user request can also be improved.

In addition, because the M customer service robots not only comprise the platform customer service robot, but also comprise a third party customer service robot, and each customer service robot is provided with service support by the corresponding service provider, so that each customer service robot is maintained by the corresponding service provider, the comprehensive service platform does not need to input a knowledge base in the whole industry and collect a large amount of data in the early stage to perform service identification training, only the service attribution of the comprehensive service platform 51 and a third-party service provider is divided, the third-party service provider inputs the knowledge content in the service range to the customer service robot, the comprehensive service platform completes the configuration of the general service intention responsible for the comprehensive service platform, the one-stop intelligent customer service realization of the comprehensive service platform can be completed quickly, so that the service support of a plurality of services can be realized quickly.

In a second aspect, as shown in fig. 5, an embodiment of the present specification further provides a system for determining a problem category based on a user request, including a client 50, a comprehensive service platform 51, and M customer service robots 52, where M is an integer greater than 1, including:

the client 50 is used for acquiring input data of a target user, generating a user request and sending the user request to the comprehensive service platform 51;

the integrated service platform 51 is configured to receive the user request sent by the client 50, and if a history request record corresponding to the target user exists, perform intention identification on the user request to obtain a target request intention of the user request; determining a target scene corresponding to the target request intention according to the historical request record of the target user and the target request intention; (ii) a According to the correspondence between the scene and the customer service robots, determining a target customer service robot corresponding to the target scene from the M customer service robots 52, and sending the user request to the target customer service machine; if the embedded point information carried by the user request is obtained; determining a target customer service robot corresponding to the user request from the M customer service robots 52 according to the buried point information;

the target customer service robot is used for receiving the user request sent by the comprehensive service platform 51; responding the user request to obtain a response result; and returns the response result to the client 50 through the integrated service platform 51.

Correspondingly, the integrated service platform 51 includes a platform customer service robot, and can invoke a three-party customer service robot of a service provider residing in the integrated service platform, the content of a knowledge base relied on by the three-party customer service robot is maintained by the corresponding service provider, and is used for acquiring a user request of a target user; performing intention identification on the user request to obtain an intention identification result; if the historical request record of the target user exists, determining a target scene corresponding to the user request according to the intention identification result and the historical request record of the target user; if the historical request record of the target user does not exist and the intention identification result is that the intention of the target request is identified, acquiring a target scene corresponding to the user request according to the corresponding relation between the preset request intention and the scene; if the intention identification result is that the intention of the target request is not identified and the user request carries the embedded point information of the last service page, determining a target scene corresponding to the user request according to the embedded point information; selecting the platform customer service robot or a certain three-party customer service robot as a target customer service robot according to the corresponding relation between a preset scene and the customer service robot; and sending the user request to the target customer service robot.

Accordingly, the integrated service platform 51 is configured to obtain a user request of a target user; performing intention identification on the user request to obtain an intention identification result; if the historical request record of the target user exists, determining a target scene corresponding to the user request according to the historical request record of the target user and the target request intention; if the historical request record of the target user does not exist and the intention identification result is that the intention of the target request is identified, acquiring a target scene corresponding to the user request according to the corresponding relation between the preset request intention and the scene; if the intention identification result is that the intention of the target request is not identified and the user request carries the embedded point information of the last service page, determining a target scene corresponding to the user request according to the embedded point information; and determining a target customer service robot corresponding to the target scene from M preset customer service robots according to the corresponding relation between the preset scene and the customer service robots, and sending the user request to the target customer service robot, wherein M is an integer greater than 1.

In the embodiment of this specification, the integrated service platform 51 provides a uniform intelligent customer service entrance for all services to live in, and a service provider entering the integrated service platform 51 needs to provide its own call information of the customer service robot, that is, each of the M customer service robots 52 is provided with the call information of the customer service robot by the corresponding service provider, where the call information is parameters required for accessing a question-and-answer interface of the customer service robot: including the robot number (usually a unique identifier), the signature verification, and the key information required for encryption (specifically for securing communication), so that the integrated service platform 51 can invoke each service robot according to the invocation information of each service robot to process the user request.

In an embodiment of the present specification, the M customer service robots include a platform customer service robot of the integrated service platform and a third party customer service robot provided by a third party service provider.

In the embodiment of the present specification, a user will carry the embedded point information of the previous service page after entering from the customer service entrance provided by the integrated service platform 51 and then access the intelligent customer service end interface (i.e. platform page) provided by the integrated service platform 51; therefore, the user request can carry the embedded point information of the previous service page, wherein the embedded point information comprises the service subject information of the previous service page.

As shown in fig. 6, the integrated service platform 51 includes a platform intelligent customer service end 53 and an intelligent routing system 54, the integrated service platform 51 provides a plurality of customer service entrances, the plurality of customer service portals include a equity services portal 61, a social security services portal 62, and other three-party services portal 63, so that a user may enter the platform intelligent customer service provider 53 through each of the plurality of customer service portals, is used for accessing the platform intelligent customer service end 53, so that the user request of the user accessing the platform intelligent customer service end 53 enters the intelligent routing system 54 after passing through the platform intelligent customer service end 53, after entering the customer service entrance, the user can carry the intelligent customer service end interface provided by the intelligent customer service end 53 of the embedded point information access platform of the previous service page.

In the embodiment of the present specification, as shown in fig. 6, the content of the knowledge base corresponding to each customer service robot in the M customer service robots 52 is maintained by each service provider, the M customer service robots 52 include a platform customer service robot 520, a public deposit customer service robot 521, a social security customer service robot 522, and a three-party customer service robot 523, and the platform customer service robot 520 is maintained by the integrated service platform 51, and can provide general support for the user about the capability and service of the platform itself.

Thus, the user request of the user access platform intelligent customer service end 53 carries the embedded point information of the last service page; and receiving the user request at the intelligent routing system 54, the intelligent routing system 54 comprehensively judging the target customer service robot of the user request according to the information of the embedded point information, the intention identification result, the user history callback data, the platform knowledge matching result and the like of the previous service page; the user request is then sent to the target customer service robot.

The intelligent routing system 54 receives the response result returned by the target customer service robot, analyzes the response result, and pushes the response result to the platform intelligent customer service server 53 when knowledge or recommended content exists; when no answer exists, the request is forwarded to the platform customer service robot again, and the platform customer service robot carries out bottom-binding processing; when the platform customer service robot performs bottom-in-process processing, the processing process data is reserved, the comprehensive service platform 51 performs data analysis periodically, the service provider is promoted to perfect a knowledge base or optimize a routing strategy according to specific reasons, and the overall customer service answer rate of the comprehensive service platform 51 is promoted to be improved.

As shown in fig. 7, a target user 70 may enter a unified customer service interface 531 provided by a platform intelligent customer service end 53 through a platform page 53 or a service page 71, where the platform page 53, the service page 71, and the unified customer service interface 531 are all displayed on a display screen of a client 50; the target user 70 generates a corresponding user request through the input data in the unified customer service interface 531, the user request firstly enters a session management unit 540 in the intelligent routing system 54, and the session management unit 540 searches the session flow water meter according to the user identifier sessionId of the user request.

The intelligent routing system 54 includes a session management unit 540, and the session management unit 540 first receives the user request sent by the client 50, and searches the session flow water meter according to the user identifier sessionId of the user request according to a history session loading module 5401 in the session management unit 540; if the history request record is found, loading the history session, and at this time, only loading the found history request record, or loading the history routing data and scene configuration in the history request record, and taking the history routing data and scene configuration into a subsequent module as a routing judgment basis.

The session management unit 540 further includes a session recording module 5402, configured to perform session flow record processing after each customer service robot successfully calls and before returning to the customer service end 50; the recorded content comprises user request information, a subsequent module processing result and response content of the robot, and is associated to sessionId, so that multiplexing of user requests in subsequent rounds is facilitated.

Further, the intelligent routing system 54 further includes an algorithm engine unit 541, configured to perform intent recognition on the user request through an intent recognition module 5411 in the algorithm engine unit 541 after the history session loading module 5401 detects that there is a history request record and loads the history request record, and determine a service scenario according to the history routing data and the scenario configuration. Specifically, when the identified intention hits the historical intention corresponding to the historical routing data, the service scene corresponding to the historical intention is used as the routing result of the request. Subsequent robot calls are made based on the content of the service scenario configuration. For example, the abnormal intention of the account hit by the questioning content of the user is associated with the scene of platform service, and even if the user enters from the Hangzhou public accumulation fund service page, the request can be routed to the platform robot for service support; the intentions configured in the algorithm engine 514 and the service scenarios associated with the intentions can be iterated and optimized continuously with the upgrading of the platform and the expansion of the service content.

The intelligent routing system 54 further includes a scene policy engine unit 542, configured to, after the session management unit 540 does not find the history request record, if the intention identification module 5411 identifies an intention of the target request corresponding to the user request, load, according to the scene configuration loading module 5422 in the scene policy engine unit 542, relevant service information configured for the target scene, where the relevant service information includes a call parameter of a customer service robot associated with the target scene, and manual and bottom-to-bottom policies of the target scene, and the like; and sends the acquired related service information to the robot routing unit 543 in the scene policy engine unit 542.

If the historical request record is not found by the session management unit 540 and the intention identification module 5411 does not identify the intention of the target request corresponding to the user request, analyzing the embedded point information of the previous service page carried by the user request according to the embedded point analysis module 5421 in the scene policy engine unit 542 to obtain a corresponding target scene, and loading the relevant service information configured by the target scene by the scene configuration loading module 5422; and sends the acquired related service information to the robot routing unit 543 in the scene policy engine unit 542.

And after receiving the related service information sent by the scene policy engine unit 542, the robot routing unit 543 packages the invocation information of the target customer service robot and the user request according to the robot request packaging module 5430 in the robot routing unit 543, and sends the packaged invocation information and the user request to the target customer service robot, where the target customer service robot may be any one of the M customer service robots 52.

Further, if the history request record is not found, the session management unit 540 identifies a question intention of the user by using the NLU through the intention identification module 5411, and if a target request intention corresponding to the user request is not identified, acquires a service scene corresponding to the previous service page according to the embedded point information of the previous service page carried by the user request, and takes the service scene as the target scene; and acquiring the customer service robot corresponding to the service scene according to the corresponding relation between the scene and the customer service robot, and sending the user request to the customer service robot corresponding to the service scene.

In addition, the robot routing unit 543 further includes an analysis unit 5431, configured to encapsulate differences of the request packets of each customer service robot and perform standardized analysis on the response packets; and the scenario policy engine unit 542 may further include a traffic optimization processing module 5423.

Further, the scenario policy engine unit 542 may further include a service optimization processing module 5423, configured to perform service optimization processing on two response situations, namely "manual transfer" and "no answer", after determining the standardized response of the target customer service robot; and converting manual response optimization logic, acquiring a scene policy configuration and a preposed robot routing result, and packaging into a clickable service card to be rendered at a client according to a manual channel (an online workbench, a hotline telephone and the like) which can be supported by a corresponding service provider (a service providing mechanism or a platform) in the scene, so that a user can quickly enter the subsequent step of the manual channel after clicking. And the non-answer response optimization logic is combined with the routing result and the scene configuration, and the non-answer response is replaced by using an artificial bottom-pocket or a worksheet bottom-pocket, the artificial bottom-pocket processing logic is the same as a, the worksheet bottom-pocket directly returns a user-friendly default text prompt, and then a non-answer worksheet is created to perform data analysis and service optimization on operators in the corresponding scene.

Therefore, the comprehensive service platform 51 does not need to input a knowledge base in the whole industry, a large amount of data does not need to be collected in an early stage to perform service identification training, only the service affiliation of the comprehensive service platform 51 and a service provider is divided, a third-party service provider inputs the knowledge content in the service range to the customer service robot of the third-party service provider, the comprehensive service platform 51 completes the configuration of the general service intention responsible for the comprehensive service platform 51, and the one-stop intelligent customer service implementation of the comprehensive service platform 51 can be rapidly completed, so that the service support of a plurality of services can be rapidly realized.

In the embodiment of the present specification, an intelligent routing system 54 is introduced for uniformly receiving access requests of all users in the integrated service platform 51, and a target customer service robot is determined based on various technical means such as a history request record, a point burying technology, an intention identification module, and the like, so that the accuracy of the identified target customer service robot is higher; the implementation idea is summarized as follows: the integrated service platform 51 provides a unified intelligent customer service entrance to receive customer requests such as customer complaints and consultation of all users. When entering a platform intelligent customer service page from a service page, requesting to carry embedded point information, and when initiating a request to a service end, an intelligent customer service client transmits upstream page information carried by the embedded point information and a login sessionID of a user; the intelligent routing system 54 performs pre-knowledge filtering according to the intention recognition model, filters common problems supported by the platform and transfers the common problems to the platform customer service robot for bearing, performs secondary service routing according to the embedded point information and the user question content, and selects the correct customer service robot, so that the accuracy of the determined target customer service robot is improved, namely the matching degree between the identified target customer service robot and the request is higher.

In addition, based on the response content and the custom configuration of each service scene, the service optimization processing can be carried out on response results such as manual response and no answer, the corresponding manual channel is directly presented to the user through a service card without self search of the user, the response time can be effectively shortened through the optimized response steps, the response efficiency is improved, and the user experience is better.

The comprehensive service platform in the embodiment of the specification can provide a one-stop handling channel for a user to handle various matters such as social security, tax, accumulated fund, water, electricity, coal, medical treatment and the like through unified entry, such as city service sections of Citizen cloud, Guangdong province, Paibao and various bank apps.

In a third aspect, based on the same technical concept, an embodiment of the present specification provides an apparatus for determining a problem category based on a user request, as shown in fig. 8, including:

a user request acquiring unit 801 configured to acquire a user request of a target user;

a target customer service robot determining unit 802, configured to perform intent recognition on the user request if there is a history request record of the target user, and obtain a target request intent of the user request; determining a target scene corresponding to the target request intention according to the historical request record of the target user and the target request intention; determining a target customer service robot corresponding to the target scene from M preset customer service robots according to the corresponding relation between the scene and the customer service robots, wherein M is an integer greater than 1; if the historical request record of the target user does not exist, determining a target customer service robot corresponding to the user request from the M customer service robots according to at least one data of the target request intention and the embedded point information of the previous service page, wherein the embedded point information of the previous service page is carried in the user request

A user request sending unit 803, configured to send the user request to the target customer service robot.

In an optional implementation manner, the target customer service robot determining unit 802 is configured to obtain a correspondence between a history request intention of the target user and a history session scene according to a history request record of the target user; detecting whether an intention which is the same as the target request intention exists in the historical request intentions, and if so, acquiring the target scene according to the corresponding relation between the historical request intention and a historical conversation scene; and if not, acquiring the target scene according to a preset scene corresponding table and the target request intention.

In an optional implementation manner, the target customer service robot determining unit 802 is configured to perform intent recognition on the user request by using a trained natural language understanding model, and obtain the target request intent.

In an alternative embodiment, the target customer service robot determining unit 802 is configured to perform intent recognition on the user request; if the target request intention is identified, determining the target scene according to the scene mapping table and the target request intention; and determining a target customer service robot corresponding to the target scene according to the corresponding relation between the scene and the customer service robot, and sending the user request to the target customer service robot.

In an optional implementation manner, the target customer service robot determining unit 802 is configured to analyze the user request to obtain the embedded point information of the previous service page, obtain a service scenario corresponding to the previous service page, and use the service scenario as the target scenario; and acquiring the customer service robot corresponding to the service scene from the M customer service robots according to the corresponding relation between the scene and the customer service robots, taking the acquired customer service robot as the target customer service robot, and sending the user request to the target customer service robot.

In an alternative embodiment, the M customer service robots include a platform customer service robot and a third party customer service robot of the integrated service platform.

In a fourth aspect, based on the same technical concept, an embodiment of the present specification provides a problem category determining apparatus based on a user request, which is applied to an integrated service platform, where the integrated service platform includes a platform customer service robot and can invoke a three-party customer service robot of a service provider hosted in the integrated service platform, and a knowledge base content relied on by the three-party customer service robot is maintained by a corresponding service provider, and includes:

In a fifth aspect, based on the same technical concept, an embodiment of the present specification provides an issue category determination apparatus based on a user request, including:

In a sixth aspect, based on the same inventive concept as the problem category determination method based on the user request in the foregoing embodiment, an embodiment of the present specification further provides an electronic device, as shown in fig. 9, including a memory 904, a processor 902, and a computer program stored on the memory 904 and executable on the processor 902, where the processor 902 implements the steps of any one of the foregoing problem category determination methods based on the user request when executing the program.

Where in fig. 9 a bus architecture (represented by bus 900), bus 900 may include any number of interconnected buses and bridges, and bus 900 links together various circuits including one or more processors, represented by processor 902, and memory, represented by memory 904. The bus 900 may also link together various other circuits such as peripherals, voltage regulators, power management circuits, and the like, which are well known in the art, and therefore, will not be described any further herein. A bus interface 905 provides an interface between the bus 900 and the receiver 901 and transmitter 903. The receiver 901 and the transmitter 903 may be the same element, i.e., a transceiver, providing a means for communicating with various other apparatus over a transmission medium. The processor 902 is responsible for managing the bus 900 and general processing, and the memory 904 may be used for storing data used by the processor 902 in performing operations.

In a seventh aspect, based on the inventive concept of the problem category determination method based on the user request in the foregoing embodiments, the present specification embodiment further provides a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, implements the steps of any one of the foregoing problem category determination methods based on the user request.

The description has been presented with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the description. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program 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 computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program 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.

While preferred embodiments of the present specification have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all changes and modifications that fall within the scope of the specification.

It will be apparent to those skilled in the art that various changes and modifications may be made in the present specification without departing from the spirit and scope of the specification. Thus, if such modifications and variations of the present specification fall within the scope of the claims of the present specification and their equivalents, the specification is intended to include such modifications and variations.

Claims

1. A method for problem category determination based on a user request, comprising:

acquiring a user request of a target user;

if the historical request record of the target user exists, performing intention identification on the user request to obtain the intention of the target request of the user request; determining a target scene corresponding to the target request intention according to the historical request record of the target user and the target request intention; according to the corresponding relation between a scene and customer service robots, determining a target customer service robot corresponding to the target scene from M preset customer service robots, and sending the user request to the target customer service robot, wherein M is an integer greater than 1;

if the historical request record of the target user does not exist, determining a target customer service robot corresponding to the user request from the M customer service robots according to at least one data of a target request intention and the embedded point information of the previous service page, and sending the user request to the target customer service robot, wherein the embedded point information of the previous service page is carried in the user request, and the target request intention is obtained by performing intention identification on the user request.

2. The determination method according to claim 1, wherein the determining a target scene corresponding to the target request intention according to the target user's history request record and the target request intention comprises:

acquiring a corresponding relation between the history request intention of the target user and a history conversation scene according to the history request record of the target user;

detecting whether an intention which is the same as the target request intention exists in the historical request intentions, and if so, acquiring the target scene according to the corresponding relation between the historical request intention and a historical conversation scene; and if not, acquiring the target scene according to a preset scene corresponding table and the target request intention.

3. The determination method according to claim 2, wherein the performing intent recognition on the user request to obtain the target request intent of the user request comprises:

and performing intention recognition on the user request by using a trained natural language understanding model to acquire the target request intention.

4. The determination method according to claim 3, wherein the determining a target service robot corresponding to the user request from the M service robots based on at least one of the target request intention and the buried point information of the previous service page comprises:

performing intention recognition on the user request;

if the target request intention is identified, determining the target scene according to the scene mapping table and the target request intention;

and determining a target customer service robot corresponding to the target scene according to the corresponding relation between the scene and the customer service robot, and sending the user request to the target customer service robot.

5. The determination method according to claim 3, wherein the determining a target service robot corresponding to the user request from the M service robots based on at least one of the target request intention and the buried point information of the previous service page comprises:

analyzing the user request to obtain the embedded point information of the previous service page; acquiring a service scene corresponding to the previous service page according to the embedded point information, and taking the service scene as the target scene;

and acquiring the customer service robot corresponding to the service scene from the M customer service robots according to the corresponding relation between the scene and the customer service robots, taking the acquired customer service robot as the target customer service robot, and sending the user request to the target customer service robot.

6. The determination method of any one of claims 1-5, said M customer service robots comprising a platform customer service robot and a third party customer service robot of said integrated services platform.

7. A problem category determination method based on user requests is applied to an integrated service platform, the integrated service platform comprises platform customer service robots and can call three-party customer service robots of service providers residing in the integrated service platform, the three-party customer service robots rely on knowledge base contents maintained by the corresponding service providers, and the method comprises the following steps:

acquiring a user request of a target user;

8. A problem category determination method based on user requests comprises the following steps:

acquiring a user request of a target user;

9. An issue category determination apparatus based on a user request, comprising:

10. The device of claim 9, wherein the target customer service robot determination unit is configured to obtain a correspondence between a history request intention of the target user and a history session scene according to a history request record of the target user; detecting whether an intention which is the same as the target request intention exists in the historical request intentions, and if so, acquiring the target scene according to the corresponding relation between the historical request intention and a historical conversation scene; and if not, acquiring the target scene according to a preset scene corresponding table and the target request intention.

11. The apparatus of claim 10, wherein the target customer service robot determination unit is configured to perform intent recognition on the user request using a trained natural language understanding model, and obtain the target request intent.

12. The apparatus of claim 11, said target customer service robot determination unit to identify intent of said user request; if the target request intention is identified, determining the target scene according to the scene mapping table and the target request intention; and determining a target customer service robot corresponding to the target scene according to the corresponding relation between the scene and the customer service robot, and sending the user request to the target customer service robot.

13. The apparatus according to claim 12, wherein the target customer service robot determining unit is configured to parse the user request to obtain the embedded point information of the previous service page, obtain a service scenario corresponding to the previous service page, and use the service scenario as the target scenario; and acquiring the customer service robot corresponding to the service scene from the M customer service robots according to the corresponding relation between the scene and the customer service robots, taking the acquired customer service robot as the target customer service robot, and sending the user request to the target customer service robot.

14. The apparatus of any one of claims 9-13, the M customer service robots to include a platform customer service robot and a third party customer service robot of the integrated services platform.

15. A problem category determination device based on user request is applied to an integrated service platform, the integrated service platform comprises a platform customer service robot and can call a three-party customer service robot of a service provider residing in the integrated service platform, the three-party customer service robot depends on knowledge base contents maintained by the corresponding service provider, and the problem category determination device comprises:

16. An issue category determination apparatus based on a user request, comprising:

17. A problem category determination system based on user requests comprises a client, a comprehensive service platform and M customer service robots, wherein M is an integer greater than 1, and the problem category determination system comprises:

18. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the method of any one of claims 1-8 when executing the program.

19. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 8.