CN114548092A

CN114548092A - Customer service session scheduling method and device, equipment, medium and product thereof

Info

Publication number: CN114548092A
Application number: CN202210172424.0A
Authority: CN
Inventors: 吴培浩
Original assignee: Guangzhou Huaduo Network Technology Co Ltd
Current assignee: Guangzhou Huaduo Network Technology Co Ltd
Priority date: 2022-02-24
Filing date: 2022-02-24
Publication date: 2022-05-27

Abstract

The application discloses a customer service session scheduling method and a device, equipment, medium and product thereof, wherein the method comprises the following steps: acquiring dialogue data of any session in the intelligent customer service system, wherein the dialogue data comprises a question text of a question user and a reply text of the system; extracting deep semantic information of a word vector corresponding to each participle from the dialogue data to obtain a corresponding semantic vector; calculating a correlation matrix between the label vector of each label in the label library and each semantic vector, and determining a label corresponding to the label vector of which the full semantic vector meets a preset correlation matching condition as a hit label according to the correlation matrix; and accessing the session with the determined hit label to the agent interface corresponding to the hit label so as to continue the session by the agent user of the agent interface and the questioning user of the session. Therefore, the shunting scheduling efficiency of the intelligent customer service system is improved, waiting time is saved for questioning users, and cost reduction and efficiency improvement are realized for merchant users.

Description

Customer service session scheduling method and device, equipment, medium and product thereof

Technical Field

The present application relates to the field of intelligent customer service technologies, and in particular, to a customer service session scheduling method, and a corresponding apparatus, computer device, computer-readable storage medium, and computer program product.

Background

In the e-commerce platform, a consumer user often performs pre-sale consultation, in-sale inquiry, after-sale feedback and the like with a merchant user through a third-party chat tool or a chat tool built in the e-commerce platform, an intelligent customer service system of the e-commerce platform is responsible for responding to a problem proposed by the consumer user, the intelligent customer service system is generally used for responding to two types of modes of an intelligent robot and a seat user, and the intelligent customer service system is specifically scheduled according to actual conditions.

Because of the large number of instant chats, the types of questions vary from customer to customer, and the importance/urgency of the store varies. In general, the agent user can answer the chat list in turn according to the sequence of the chat list or answer the question according to the subjective intention of the agent. It is difficult to efficiently order and prioritize the questions because the content and type of the customer questions cannot be known in advance. When a merchant has a plurality of different seats, and the different seats are responsible for solving problems with different contents, the system can not automatically shunt for conversation, manual shunting treatment needs to be carried out by an administrator, and time and labor are wasted.

In the existing scheme, generally, a sorting function is added to a system to screen conversations according to time, so that the waiting time of customers is reduced, or manual screening is performed based on keywords, the processing mode is low in efficiency, intelligent shunting scheduling processing on conversations cannot be realized, and the waiting time of consumer users cannot be really and effectively reduced, so that improvement is needed.

Disclosure of Invention

A primary object of the present application is to solve at least one of the above problems and provide a method for scheduling a customer service session, and a corresponding apparatus, computer device, computer readable storage medium, and computer program product.

In order to meet various purposes of the application, the following technical scheme is adopted in the application:

the customer service session scheduling method adapted to one of the purposes of the present application includes the following steps:

acquiring dialogue data of any session in an intelligent customer service system, wherein the dialogue data comprises a question text generated by a question user of the session and a reply text generated by the system;

extracting deep semantic information of word vectors corresponding to each participle from the dialogue data to obtain semantic vectors corresponding to each word vector;

calculating a correlation matrix between a label vector corresponding to each label in a preset label library and each semantic vector, and determining a label corresponding to a label vector of which the full semantic vector meets a preset correlation matching condition as a hit label according to the correlation matrix;

and accessing the session corresponding to the dialogue data with the determined hit label to the seat interface corresponding to the hit label, so that the seat user corresponding to the seat interface and the questioning user of the session continue the session.

In a deepened embodiment, the method for extracting deep semantic information of the word vector corresponding to each participle from the dialogue data to obtain the semantic vector corresponding to each word vector includes the following steps:

sequentially splicing the question text and the reply text in the dialogue data according to the created time sequence to form a dialogue text;

carrying out preset standardized preprocessing on the dialog text to enable the dialog text to become a standard text;

performing word segmentation processing on the standard text to obtain a word segmentation sequence of the standard text;

querying a preset word vector table, determining word vectors corresponding to all the participles in the participle sequence, and obtaining a word vector sequence corresponding to the standard text;

and extracting deep semantic information of each word vector in the word vector sequence by using a text extraction model which is trained to be in a convergence state in advance and referring to the context information, and mapping the deep semantic information to a high-dimensional space to obtain a semantic vector corresponding to each word vector.

In a deepened embodiment, a correlation matrix between a tag vector corresponding to each tag in a preset tag library and each semantic vector is calculated, and a tag corresponding to a tag vector of which the full quantity of semantic vectors meet a preset correlation matching condition is determined as a hit tag according to the correlation matrix, which includes the following steps:

calculating a correlation matrix between a tag vector corresponding to each tag in a preset tag library and each semantic vector by adopting a preset correlation algorithm, wherein the correlation matrix comprises a correlation coefficient of each semantic vector mapped to each tag vector;

according to the correlation coefficient, a preset voting algorithm is applied to calculate a comprehensive correlation coefficient which is mapped to each label vector by all semantic vectors so as to represent the correlation probability of the dialogue data belonging to each label vector;

and screening according to preset relevant matching conditions, and determining one or more label vectors with the highest relevant probability as the label vector(s) most relevant to the full-amount semantic vector(s), wherein the corresponding label(s) is (are) a hit label(s).

In an extended embodiment, before the step of calculating a correlation matrix between a tag vector corresponding to each tag in a preset tag library and each semantic vector, the method includes the following steps:

acquiring label configuration information corresponding to an intelligent customer service system, wherein the label configuration information comprises mapping relation data between a label and one or more keywords of the label;

and coding each keyword in the tag configuration information by adopting a tag coding model which is trained to be in a convergence state in advance to obtain a tag vector corresponding to the corresponding deep semantic information, and storing the tag vector in the tag library.

In a further embodiment, accessing the session corresponding to the dialog data for which the hit tag is determined to the agent interface corresponding to the hit tag includes the following steps:

inquiring a label tag corresponding to each preset seat interface, and determining the seat interface of which the label tag completely comprises all hit labels as the seat interface corresponding to the hit labels;

establishing a data communication link between the determined agent interface and a questioning user of the session corresponding to the dialogue data of which the hit tag is determined, so that the session is continued by the agent user corresponding to the agent interface and the questioning user of the session;

and pushing a notification message representing the switching agent user to a chat interface at the questioning user side corresponding to the session.

In the expanded embodiment, after the step of determining, according to the correlation matrix, a tag corresponding to a tag vector whose full-size semantic vector meets a preset correlation matching condition as a hit tag, the method includes the following steps:

responding to a session management request of any agent user, and pushing a session list to the agent user, wherein the session list comprises all ongoing sessions of the intelligent customer service system and mapping relation data between the ongoing sessions and the hit labels of the ongoing sessions;

receiving a hit tag appointed by the agent user, updating the session list, and enabling the session list to only contain the session corresponding to the appointed hit tag;

and responding to a participation request of the agent user selecting any one session in the session list, and establishing a data communication link between the agent user and a questioning user of the session so as to continue the session by the agent user and the questioning user.

The present application provides a customer service session scheduling device adapted to one of the purposes of the present application, including: the system comprises a data acquisition module, a semantic extraction module, a label determination module and a communication establishment module, wherein the data acquisition module is used for acquiring dialogue data of any session in the intelligent customer service system, and the dialogue data comprises a question text generated by a question user of the session and a reply text generated by the system; the semantic extraction module is used for extracting deep semantic information of word vectors corresponding to each participle from the dialogue data to obtain semantic vectors corresponding to each word vector; the tag determining module is used for calculating a correlation matrix between a tag vector corresponding to each tag in a preset tag library and each semantic vector, and determining tags corresponding to tag vectors which satisfy a preset correlation matching condition with the full amount of semantic vectors as hit tags according to the correlation matrix; and the communication establishing module is used for accessing the session corresponding to the dialogue data with the determined hit label to the seat interface corresponding to the hit label so that the seat user corresponding to the seat interface and the questioning user of the session continue the session.

In a further embodiment, the semantic extraction module includes: the text splicing submodule is used for sequentially splicing the question text and the reply text in the dialogue data according to the created time sequence to form a dialogue text; the text specification submodule is used for carrying out preset standardized preprocessing on the conversation text to enable the conversation text to become a specification text; the text word segmentation sub-module is used for carrying out word segmentation processing on the standard text to obtain a word segmentation sequence of the standard text; the word segmentation coding sub-module is used for querying a preset word vector table, determining word vectors corresponding to all the segmentation words in the word segmentation sequence and obtaining a word vector sequence corresponding to the standard text; and the semantic extraction submodule is used for extracting deep semantic information of each word vector in the word vector sequence by using a text extraction model which is trained to be in a convergence state in advance and referring to the context information, and mapping the deep semantic information to a high-dimensional space to obtain a semantic vector corresponding to each word vector.

In a further embodiment, the tag determination module includes: the matrix construction submodule is used for calculating a correlation matrix between a label vector corresponding to each label in a preset label library and each semantic vector by adopting a preset correlation algorithm, and the correlation matrix comprises a correlation coefficient of each semantic vector mapped to each label vector; the coefficient synthesis submodule is used for calculating a synthesis correlation coefficient corresponding to all semantic vectors mapped to each label vector by applying a preset voting algorithm according to the correlation coefficient so as to represent the correlation probability of the dialogue data belonging to each label vector; and the tag hit submodule is used for screening according to preset relevant matching conditions, determining one or more tag vectors with the highest relevant probability as the tag vector(s) most relevant to the full-scale semantic vector, and taking the corresponding tag(s) as hit tags.

In an extended embodiment, the customer service session scheduling apparatus of the present application further includes: the system comprises a configuration acquisition module, a configuration processing module and a configuration processing module, wherein the configuration acquisition module is used for acquiring label configuration information corresponding to the intelligent customer service system, and the label configuration information comprises mapping relation data between a label and one or more keywords of the label; and the tag extraction module is used for coding each keyword in the tag configuration information by adopting a tag coding model which is trained to be in a convergence state in advance to obtain a tag vector corresponding to the corresponding deep semantic information, and storing the tag vector in the tag library.

In a further embodiment, the communication establishing module includes: the seat determining submodule is used for inquiring the label tags corresponding to the preset seat interfaces and determining the seat interfaces of which the label tags completely contain all hit labels as the seat interfaces corresponding to the hit labels; the link establishment submodule is used for establishing a data communication link between the determined agent interface and the questioning user of the session corresponding to the dialogue data of which the hit label is determined so that the session can be continued by the agent user corresponding to the agent interface and the questioning user of the session; and the notification pushing submodule is used for pushing a notification message representing the switching agent user to a chat interface of the questioning user side corresponding to the session.

In an extended embodiment, the customer service session scheduling apparatus of the present application further includes: the initial pushing submodule is used for responding to a session management request of any agent user and pushing a session list to the agent user, wherein the session list comprises all ongoing sessions of the intelligent customer service system and mapping relation data among the hit labels; the updating and pushing submodule is used for receiving the hit tag appointed by the agent user and updating the session list to ensure that the session list only contains the session corresponding to the appointed hit tag; and the call switching submodule is used for responding to a participation request of any one session selected by the agent user in the session list, and establishing a data communication link between the agent user and the questioning user of the session so that the agent user and the questioning user continue the session.

A computer device adapted for one of the purposes of the present application comprises a central processing unit and a memory, the central processing unit being configured to invoke the execution of a computer program stored in the memory to perform the steps of the customer service session scheduling method described herein.

A computer-readable storage medium, which stores a computer program implemented according to the method for scheduling customer service sessions in the form of computer-readable instructions, and when the computer program is called by a computer, executes the steps included in the method.

A computer program product, provided to adapt to another object of the present application, comprises computer programs/instructions which, when executed by a processor, implement the steps of the method described in any of the embodiments of the present application.

Compared with the prior art, the application has the following advantages:

firstly, aiming at a conversation in an intelligent customer service system, a correlation matrix is calculated by a semantic vector obtained by segmenting conversation data corresponding to the conversation and a preset label vector of a label, then the label meeting a preset matching condition is matched for the conversation data according to the correlation matrix, then the conversation data is distributed according to the label of the conversation data, and the conversation data is distributed to the seat interface matched with the label, so that the questioning user of the conversation establishes data communication connection with the seat interface, the seat user of the seat interface can be in the conversation with the corresponding content according to the preset label, the automatic shunting scheduling of the conversation is realized by automatically matching the seat interface for the conversation intelligence, manual distribution is not needed, the service effect of an intelligent customer service system is improved, the time length of asking users to wait for the distribution of seat users is reduced, and the user experience is improved.

Secondly, the intelligent customer service system is allowed to define the tags by setting the semantic probes through the tags which are actually played by the application, the tags can be effective in real time through configuration, the urgency, the different functions, the different products and the like of each conversation can be indicated, and the conversation can be identified in real time, so that the seat user and the conversation can be matched in a way of being related to the tags, the phenomenon of strong rule matching completely depending on keywords in the traditional technology is avoided, the conversation with similar semantics but not containing the keywords can be recalled better, and the intelligent customer service system is more Robust in operation (Robust).

In addition, by implementing the technical scheme, a large amount of manual work can be omitted in large customer service scenes such as e-commerce platforms, manual browsing conversations of managers of users of various merchants are omitted, corresponding implementation cost is saved, and large-scale economic utility is obtained.

Drawings

The above and/or additional aspects and advantages of the present application will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:

FIG. 1 is a schematic flow chart diagram illustrating an exemplary embodiment of a customer service session scheduling method according to the present application;

FIG. 2 is a schematic flow chart of extracting semantic vectors for the dialogue data in the embodiment of the present application;

FIG. 3 is a schematic flow chart illustrating a semantic vector and tag vector matching process according to an embodiment of the present disclosure;

FIG. 4 is a flow chart illustrating a tag configuration process according to an embodiment of the present application;

FIG. 5 is a flowchart illustrating a process for scheduling agent interfaces for sessions in an embodiment of the present application;

FIG. 6 is a flowchart illustrating a process of screening sessions by an agent user in an embodiment of the present application;

FIG. 7 is a functional block diagram of a customer service session scheduler according to the present application;

fig. 8 is a schematic structural diagram of a computer device used in the present application.

Detailed Description

Reference will now be made in detail to embodiments of the present application, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are exemplary only for the purpose of explaining the present application and are not to be construed as limiting the present application.

As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may also be present. Further, "connected" or "coupled" as used herein may include wirelessly connected or wirelessly coupled. As used herein, the term "and/or" includes all or any element and all combinations of one or more of the associated listed items.

It will be understood by those within the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.

As will be appreciated by those skilled in the art, "client," "terminal," and "terminal device" as used herein include both devices that are wireless signal receivers, which are devices having only wireless signal receivers without transmit capability, and devices that are receive and transmit hardware, which have receive and transmit hardware capable of two-way communication over a two-way communication link. Such a device may include: cellular or other communication devices such as personal computers, tablets, etc. having single or multi-line displays or cellular or other communication devices without multi-line displays; PCS (Personal Communications Service), which may combine voice, data processing, facsimile and/or data communication capabilities; a PDA (Personal Digital Assistant), which may include a radio frequency receiver, a pager, internet/intranet access, a web browser, a notepad, a calendar and/or a GPS (Global Positioning System) receiver; a conventional laptop and/or palmtop computer or other device having and/or including a radio frequency receiver. As used herein, a "client," "terminal device" can be portable, transportable, installed in a vehicle (aeronautical, maritime, and/or land-based), or situated and/or configured to operate locally and/or in a distributed fashion at any other location(s) on earth and/or in space. The "client", "terminal Device" used herein may also be a communication terminal, a web terminal, a music/video playing terminal, such as a PDA, an MID (Mobile Internet Device) and/or a Mobile phone with music/video playing function, and may also be a smart tv, a set-top box, and the like.

The hardware referred to by the names "server", "client", "service node", etc. is essentially an electronic device with the performance of a personal computer, and is a hardware device having necessary components disclosed by the von neumann principle such as a central processing unit (including an arithmetic unit and a controller), a memory, an input device, an output device, etc., a computer program is stored in the memory, and the central processing unit calls a program stored in an external memory into the internal memory to run, executes instructions in the program, and interacts with the input and output devices, thereby completing a specific function.

It should be noted that the concept of "server" as referred to in this application can be extended to the case of a server cluster. According to the network deployment principle understood by those skilled in the art, the servers should be logically divided, and in physical space, the servers may be independent from each other but can be called through an interface, or may be integrated into one physical computer or a set of computer clusters. Those skilled in the art will appreciate this variation and should not be so limited as to restrict the implementation of the network deployment of the present application.

One or more technical features of the present application, unless expressly specified otherwise, may be deployed to a server for implementation by a client remotely invoking an online service interface provided by a capture server for access, or may be deployed directly and run on the client for access.

Unless specified in clear text, the neural network model referred to or possibly referred to in the application can be deployed in a remote server and used for remote call at a client, and can also be deployed in a client with qualified equipment capability for direct call.

Various data referred to in the present application may be stored in a server remotely or in a local terminal device unless specified in the clear text, as long as the data is suitable for being called by the technical solution of the present application.

The person skilled in the art will know this: although the various methods of the present application are described based on the same concept so as to be common to each other, they may be independently performed unless otherwise specified. In the same way, for each embodiment disclosed in the present application, it is proposed based on the same inventive concept, and therefore, concepts of the same expression and concepts of which expressions are different but are appropriately changed only for convenience should be equally understood.

The embodiments to be disclosed herein can be flexibly constructed by cross-linking related technical features of the embodiments unless the mutual exclusion relationship between the related technical features is stated in the clear text, as long as the combination does not depart from the inventive spirit of the present application and can meet the needs of the prior art or solve the deficiencies of the prior art. Those skilled in the art will appreciate variations therefrom.

The customer service session scheduling method can be programmed into a computer program product, is deployed in a client or a server to run, and is generally deployed in the server to implement, for example, in an e-commerce platform application scenario of the present application, so that the method can be executed by accessing an open interface after the computer program product runs and performing human-computer interaction with a process of the computer program product through a graphical user interface.

An exemplary application scenario of the application is an application in an e-commerce platform based on independent stations, each independent station is a business instance of the e-commerce platform, and has an independent access domain name, and an actual owner of the business instance is responsible for issuing and updating commodities.

The merchant instance of each independent station can be configured with an intelligent customer service system provided by an e-commerce platform to realize the introduction of an intelligent customer service robot, the intelligent customer service system is used for providing consultation service for a related consumer questioning user, a corresponding session is established, the questioning user enters a chat interface corresponding to the session and inputs questions needing consultation as questioning texts, after the intelligent customer service system of the e-commerce platform receives the questioning texts, under one condition, the intelligent customer service robot performs semantic matching by using the questioning texts and question sets in a knowledge base pre-configured for the independent station to match preset questions most similar to the questioning texts, standard questions in a question set to which the intelligent customer service robot belongs are determined according to the preset questions, then a pre-stored answer set mapped with the standard questions is called to determine one preset answer as a target answer, and outputting the text to the chat interface as a reply text so as to respond to the questions of the consumer user and meet the consultation requirements of the consumer user. In another case, the conversation is switched to an agent interface, so that the questioning user and the agent user realize direct manual conversation, and the agent user gives a corresponding reply text for the questioning text submitted by the questioning user. Under the two conditions, the intelligent customer service system can perform switching scheduling according to specific conditions. The two situations can be mutually switched by the support of the intelligent customer service system, and any one situation can be configured as the default situation of the intelligent customer service system.

For the latter case, when there are multiple said agent interfaces, that is, there are multiple corresponding agent users, and when the corresponding agent interfaces are allocated for the sessions, the sessions are usually required to be shunted according to a certain method, so that the sessions with different characteristics are correspondingly scheduled to different agent interfaces, and the different agent users are responsible for answering user questions. Thus, for the latter case of manual reply listed above, the dispatch session is one of the functions that need to be implemented in this application scenario.

Referring to the application scenario disclosed above, referring to fig. 1, the method for scheduling a customer service session in the present application, in an exemplary embodiment thereof, includes the following steps:

step S1300, obtaining dialogue data of any session in the intelligent customer service system, wherein the dialogue data comprises a question text generated by a question user of the session and a reply text generated by the system:

for each session, the chat records generated by the interaction between the questioning user and the robot or the agent user in the chat interface, including the questioning text submitted by the questioning user and the reply text automatically replied by the agent user or the robot, are usually stored in a database, and constitute the corresponding session data of the session.

Generally speaking, the intelligent customer service system will concurrently respond to the session requests of different questioning users to create a corresponding session for each questioning user, so the intelligent customer service system will generally maintain a plurality of said sessions simultaneously, and each session will generate said session data.

When a customer user who logs in at a client device enters a chat interface of the intelligent customer service system for the first time, the intelligent customer service system generally calls the intelligent robot first to make a first response so as to send welcome information and the like to the chat interface, and the customer user, namely a questioning user, inputs related questioning texts to the chat interface so as to obtain corresponding reply texts.

The dialogue data may be a chat record corresponding to the full amount of history data of the corresponding session, or may be a chat record corresponding to the current question and answer period (generally, the current day), which may be flexibly determined by those skilled in the art.

Step S1400, extracting deep semantic information of word vectors corresponding to each participle from the dialogue data to obtain semantic vectors corresponding to each word vector:

firstly, the dialogue data is used as a complete text, word segmentation is carried out after preset standardized preprocessing to obtain a word segmentation sequence, then word segmentation is coded by adopting a preset word vector table to obtain word vectors of all the word segmentation, a corresponding word vector sequence is obtained, deep semantic information of the word vector sequence is extracted by adopting a preset neural network model, and finally semantic vectors corresponding to all the word vectors are obtained, wherein the semantic vectors are the expression of deep semantic features of the corresponding word vectors. Considering that the dialogue data itself is organized on a question-and-answer basis, the neural network model is suitable for adopting a corresponding model with context information processing capability, particularly a neural network model with a multi-head attention mechanism added. With regard to specific implementations of this step, more detailed implementations will be provided later in this application, which will be omitted from this description.

Step S1500, calculating a correlation matrix between the label vector corresponding to each label in a preset label library and each semantic vector, and determining labels corresponding to the label vectors of which the full quantity of semantic vectors meet preset correlation matching conditions as hit labels according to the correlation matrix:

the intelligent customer service system is provided with a label library, wherein a plurality of labels and mapping relation data of one or more keywords for describing the labels are stored in the label library, and the labels and the keywords are generally created, modified and deleted by a management user of the intelligent customer service system for maintenance. In a specific application example, the tags may be divided according to the severity of the involved transaction, for example, the tags include tags such as "lighter," "severe," "deferrable," "urgent," and the like, so that the keyword corresponding to each tag may be a vocabulary describing a specific service type, service scenario, and content of the transaction, for example, for the tag of "severe," the keyword may be described as "bad," "angry," "quality problem," "defect," "return," and the like. Similarly, for the "urgency tag," its keywords may correspond to "return", "not buy", "due", etc. It can be seen that the keywords in the tag library are used to describe and define the role of a tag, and the tag itself can serve as the type label. The person skilled in the art sets the type division standard by himself, sets a plurality of labels, and uses the corresponding keywords of the labels to describe and define the labels accordingly. Based on the principle, in another embodiment, the tags may also be divided according to business links related to e-commerce orders, for example, according to different stages of "before sale", "in sale", and "after sale", so as to facilitate management of business links according to orders for seat users. And so on, as may be flexibly implemented by those skilled in the art.

For the labels in the label library, a neural network model which is trained to a convergence state in advance can be adopted to extract deep semantic information from corresponding keywords and construct the deep semantic information into a label vector, and the label vector can be extracted in real time or can be extracted in advance and stored in the label library for calling. In a word, the semantic representation of the keywords of the corresponding labels can be realized through the label vectors, so that the intelligent customer service system can conveniently use the label vectors to carry out semantic matching.

For the full semantic vector determined by the dialog data of one session, a preset correlation algorithm may be adopted to perform correlation calculation on the tag vectors corresponding to the full tags, so as to obtain a correlation matrix. The correlation algorithm may be implemented by using any one of algorithms for calculating data distances, such as cosine similarity algorithm, euclidean distance, pearson correlation coefficient, and jaccard algorithm. After calculation of a correlation algorithm, a corresponding correlation coefficient can be determined between the full semantic vector and the full label vector, and between each semantic vector and each label vector, so that a normalized comprehensive correlation coefficient of the full semantic vector mapped to each label vector is calculated, one or more label vectors most relevant to the full semantic vector can be determined by adopting a preset correlation matching condition according to the comprehensive correlation coefficient of each label vector, and the labels corresponding to the label vectors are hit labels corresponding to the conversation data of the conversation.

The correlation matching condition may be preset to adopt a maximum value among all the comprehensive correlation coefficients, and thus, a tag corresponding to the maximum value is the hit tag; or may be preset as a preset threshold for screening all the comprehensive correlation coefficients, and thus, the tag corresponding to the tag vector whose comprehensive correlation coefficient is greater than the preset threshold is the hit tag.

Step S1600, accessing the session corresponding to the dialog data for which the hit tag is determined to the agent interface corresponding to the hit tag, so that the agent user corresponding to the agent interface and the questioning user of the session continue the session:

in the intelligent customer service system, each agent interface is pre-configured with the label in the label library, for example, the agent interface of an agent user is configured to associate two labels of "urgent" and "important", and then if the hit label of a session is matched with the label pre-labeled by the agent interface, the session can be allocated to the agent interface for processing.

After the session data of one session determines the hit tag corresponding to the session data, the agent interface matching with the hit tag can be found from all the agent interfaces according to a preset matching policy, which is specifically determined according to the matching policy. For example, in one embodiment, the matching policy is configured such that when the annotation tag of the agent interface completely contains all hit tags of the session, the two are considered to be matched, and thus the session can be shunted to the agent interface for processing. In another embodiment, the matching policy is configured to bypass the session to the agent interface for processing when one of the annotation tags of the agent interface is consistent with one of the hit tags of the session.

Therefore, the method and the system have a plurality of flexible configurations, are convenient for the background of the intelligent customer service system to be flexibly set, comprise the steps of flexibly setting the matching strategy, flexibly setting the relevant matching conditions and the like, and open more abundant functions for relevant merchant users configuring the intelligent customer service system.

After a session is assigned to a matching agent interface, the intelligent customer service system establishes a data communication link between the questioning user of the session and the agent user corresponding to the agent interface, and subsequently displays the reply text of the agent user on the chat interface of the questioning user, instead of displaying the automatic reply text of the robot, thereby actually switching the session from the robot automatic reply mode to the manual customer service mode.

When a plurality of the agent interfaces exist and a large number of questioning user interactions are sent in parallel, the intelligent customer service system can respectively schedule the matched agent interfaces for the sessions of the questioning users in real time through the processes of the steps, and accurate response task allocation can be realized without manual intervention, so that the method is very efficient.

From the above disclosure of the exemplary embodiment and its modified embodiments, it can be known that the technical solution of the present application has various positive effects, including but not limited to the following aspects:

firstly, aiming at a conversation in an intelligent customer service system, the application calculates a correlation matrix by a semantic vector obtained by segmenting conversation data corresponding to the conversation and a label vector of a preset label, matches the conversation data with the label meeting a preset matching condition according to the correlation matrix, then the conversation data is distributed according to the label of the conversation data, and the conversation data is distributed to the seat interface matched with the label, so that the questioning user of the conversation establishes data communication connection with the seat interface, the seat user of the seat interface can be in the conversation with the corresponding content according to the preset label, the automatic shunting scheduling of the conversation is realized by automatically matching the seat interface for the conversation intelligence, manual distribution is not needed, the service effect of an intelligent customer service system is improved, the time length of asking users to wait for the distribution of seat users is reduced, and the user experience is improved.

Referring to fig. 2, in a further embodiment, the step S1400 is to extract deep semantic information of the word vector corresponding to each participle from the dialogue data to obtain a semantic vector corresponding to each word vector, and includes the following steps:

step S1410, sequentially concatenating the question text and the reply text in the dialog data according to the created time sequence to form a dialog text:

in the intelligent customer service system, the question text and the reply text of the dialogue data are organized according to the generation time, namely the creation time, and the time actually guides the context relationship, so that the question text and the reply text in the dialogue data can be sequentially spliced into the same continuous text according to the creation time to form the dialogue text.

Step S1420, performing preset standardized preprocessing on the dialog text to make it a standard text:

then, according to a preset standardized preprocessing program, format preprocessing is performed on the dialog text, including removing stop words such as language words and auxiliary words, removing spaces, and removing punctuations such as symbols, expressions, noise information, and the like, for example, a text "why my express delivery does not go to o now" is input, after preprocessing, a text "why my express delivery does not go to o now" is obtained, and the preprocessed text is the standard text.

Step S1430, performing word segmentation processing on the standard text to obtain a word segmentation sequence of the standard text:

further, the canonical text is subjected to word segmentation processing by using a preset word segmentation model, the word segmentation model may use a statistical-based machine learning algorithm or a dictionary-based word segmentation algorithm, such as an HMM, CRF, SVM, a deep learning algorithm constructed by a basic model such as LSTM + CRF, Bert + CRF, and the like, and those skilled in the art can flexibly implement the word segmentation processing to obtain a word segmentation sequence including a full number of words in the canonical text.

Step S1440, querying a preset word vector table, determining word vectors corresponding to each participle in the participle sequence, and obtaining a word vector sequence corresponding to the standard text:

then, vector encoding is carried out on the word segmentation sequence, and each word segmentation in the word segmentation sequence can be converted into word vector representation through querying a word vector table, so that a word vector sequence corresponding to the word segmentation sequence is obtained. The Word vector table is obtained by pre-training and can be obtained by known pre-training models such as Bert, Word2Vec, GloVe, Fasttext, ELMO and the like, and can be flexibly selected by a person skilled in the art.

Step S1450, extracting deep semantic information of each word vector in the word vector sequence with reference to context information by using a text extraction model trained to a convergent state in advance, and mapping the deep semantic information to a high-dimensional space to obtain a semantic vector corresponding to each word vector:

and finally, adopting a text extraction model which is pre-selected and trained to be in a convergence state and is suitable for extracting deep semantic information of the word vector sequence, wherein the model can be constructed by basic models such as Bert, RNN, CNN and the like, particularly basic models with a multi-head attention mechanism such as Bert, Electrora, Transformer, LSTM, BilsTM and the like, so that the multi-head attention mechanism is added, the context information among word vectors can be fully referred in the semantic extraction process, and a more effective representation learning effect is obtained. For such models, sufficient training data sets are used to train them to learn the ability to extract deep semantic information from the word vector sequence and map it to a high-dimensional space, and the high-dimensional vector sequence obtained for the word vector sequence is a semantic vector sequence that includes semantic vectors corresponding to each word vector, each semantic vector being a representation of a corresponding word vector in deep semantics. Thus, deep semantic representation of the dialogue data is realized, and the matching process with the label vector can be participated subsequently.

In the embodiment, the dialogue data of the dialogue is subjected to more specific semantic processing, so that the representation of the dialogue data on deep semantics is realized, and the obtained semantic vector has more effective feature representation capability due to the fact that the context information of the dialogue data is referred to in the process of semantic extraction, and is beneficial to guiding the matching of the semantic vector and the tag vector, so that the matching accuracy is improved.

Referring to fig. 3, in a further embodiment, the step S1500 of calculating a correlation matrix between tag vectors corresponding to each tag in a preset tag library and each semantic vector, and determining, as hit tags, tags corresponding to tag vectors whose full amount of semantic vectors satisfy a preset correlation matching condition according to the correlation matrix includes the following steps:

step S1510, calculating a correlation matrix between the tag vector corresponding to each tag in the preset tag library and each semantic vector by using a preset correlation algorithm, where the correlation matrix includes a correlation coefficient of each semantic vector mapped to each tag vector:

as described above, the correlation algorithm may be implemented by using any algorithm for calculating the data distance, such as cosine similarity algorithm, euclidean distance, pearson correlation coefficient, and jaccard algorithm. For example, in this embodiment, a cosine similarity algorithm is used as a preset correlation algorithm, and a similarity value between each tag vector and each semantic vector is calculated as a correlation coefficient, so as to obtain a correlation matrix, where one dimension of the correlation matrix is used to correspondingly represent each tag vector, the other dimension of the correlation matrix is used to correspondingly represent each semantic vector, and each element correspondingly represents a correlation coefficient between its corresponding tag vector and its corresponding semantic vector.

Step S1520, according to the correlation coefficient, applying a preset voting algorithm to calculate a comprehensive correlation coefficient corresponding to all semantic vectors mapped to each of the tag vectors to represent a correlation probability that the dialog data belongs to each tag vector:

it is understood that, in the correlation matrix, for each tag vector, the correlation coefficients corresponding to each semantic vector are summarized to obtain a comprehensive correlation coefficient, so that the comprehensive correlation coefficient can indicate the correlation probability of mapping the full-scale semantic vector of the dialogue data to the tag vector. The way of summarizing may be to average or sum the correlation coefficients of the full semantic vector corresponding to the tag vector, as long as it is normalized to a standard numerical space, e.g., [0,1], for comparison with other comprehensive correlation coefficients. In practice, this may be done through a pooling layer. Therefore, a comprehensive correlation coefficient sequence corresponding to a full-scale label vector can be obtained, and each comprehensive correlation coefficient in the sequence represents the correlation probability of the dialog data belonging to the label corresponding to the current comprehensive correlation coefficient.

Step S1530, performing screening according to preset relevant matching conditions, and determining that the one or more tag vectors with the highest relevant probability are the tag vectors most relevant to the full semantic vector, and the corresponding tags are hit tags:

as mentioned above, the correlation matrix may be filtered by presetting a correlation matching condition, for example, the correlation matching condition gives a preset threshold, each comprehensive correlation coefficient in the comprehensive correlation coefficient sequence is compared with the preset threshold to reduce the comprehensive correlation coefficient below the preset threshold, only the comprehensive correlation coefficient above the preset threshold is retained, and the tags corresponding to the retained comprehensive correlation coefficients may include one or more tags, that is, the tags may be determined as the hit tags of the present application, that is, the hit tags corresponding to the dialogue data to which the full-scale semantic vector belongs. The hit tags are determined and can be used to match agent interfaces for the corresponding sessions of session data.

In the embodiment, the correlation matrix is used for matching the labels for the semantic vectors corresponding to the sessions, so that the agent interfaces can be conveniently matched for the sessions according to the labels, the calculation amount is small, the calculation is quick, and the method is particularly suitable for providing high-efficiency calculation efficiency for massive sessions in a high-concurrency scene, so that the response rate of the intelligent customer service system for allocating the agent interfaces for the sessions is improved. Different from the situation of strong rule matching depending on keywords in the prior art, the process of semantic matching based on deep semantic information can enable the intelligent customer service system to have higher robustness (Robust).

Referring to fig. 4, in an expanded embodiment, before the step of calculating a correlation matrix between a tag vector corresponding to each tag in a preset tag library and each semantic vector in the step S1500, the method includes the following steps:

step S1100, obtaining label configuration information corresponding to the intelligent customer service system, wherein the label configuration information comprises mapping relation data between a label and one or more keywords of the label:

in order to facilitate daily maintenance of a background management user of the intelligent customer service system, the intelligent customer service system opens a maintenance function of the management user at the background, and provides a configuration page for the management user to acquire tag configuration information. The label configuration information allows the management user to describe and define the labels in the label library, and the management user provides the labels and one or more keywords for describing and defining the labels through the label configuration information, so that the content in the label library is updated in time.

As mentioned above, the tags are divided according to a certain type division standard, and correspondingly, the tags can be described by one or more keywords, and mapping relationship data is formed between the tags and the keywords and submitted to the intelligent customer service system for association storage.

Step S1200, coding each keyword in the tag configuration information by adopting a tag coding model which is trained to a convergence state in advance to obtain a tag vector corresponding to corresponding deep semantic information, and storing the tag vector in the tag library:

further, a label coding model trained to a convergence state in advance is adopted to code each keyword in the label configuration information to obtain a corresponding label vector, the specific process can be that each keyword of a label is converted into a corresponding word vector by inquiring a preset word vector table to obtain a word vector sequence, then the label coding model extracts deep semantic information based on each word vector in the word vector sequence and maps the deep semantic information to a high-dimensional space to obtain a corresponding label vector, and the label vector can be stored in the label library together with the corresponding label.

Similar to the text extraction model, the tag coding model may be constructed by basic models such as Bert, RNN, CNN, and the like, or further constructed by basic models with a multi-head attention mechanism, such as Bert, electrora, Transformer, LSTM, BiLSTM, and the like. And then training the label to a convergence state through a sufficient training data set, so that the label is suitable for representing and learning the key words corresponding to the label. Since the text extraction model and the label coding model are implemented in a substantially similar manner, in an alternative embodiment, the two models may be the same neural network model, thereby further saving training costs.

It should be noted that the present embodiment does not necessarily depend on the prior execution of step S1300 and step S1400, both belonging to the concurrently executable steps.

In the embodiment, the tag configuration function is accessed to the intelligent customer service system, so that the intelligent customer service system can conveniently maintain the tag library, a merchant user can conveniently update the tags in the tag library in time, and the configuration of the tags in the tag library is technically decoupled from the calling of the tag vectors, so that the added tags can also take effect immediately, and the intelligent degree of the intelligent customer service system is improved.

Referring to fig. 5, in a further embodiment, in step S1600, accessing the session corresponding to the dialog data for which the hit tag is determined to the agent interface corresponding to the hit tag includes the following steps:

step S1610, querying a label corresponding to each preset seat interface, and determining a seat interface whose label completely contains all hit labels as a seat interface corresponding to the hit label:

as described above, the merchant user may pre-associate the labels in the label library for the agent interface corresponding to each agent user in the background of the intelligent customer service system, so that the labels become the label labels corresponding to the agent interfaces, so as to indicate the sessions corresponding to the labels that the agent users are suitable for processing.

Therefore, in order to implement the shunting of the session corresponding to the session data, it is first required to query the label tags carried by each agent interface, in this embodiment, according to a preset matching policy, when the label tags carried by one agent interface completely include all hit tags of one session, the agent interface is determined as the agent interface corresponding to the hit tags.

Step S1620, establishing a data communication link between the determined agent interface and the questioning user of the session corresponding to the dialog data for which the hit tag is determined, so that the agent user corresponding to the agent interface and the questioning user of the session continue the session:

after the agent interface is determined, the system can further distribute the session to the determined agent interface, and establish a data communication link between the agent interface and the questioning user, so that the agent user of the agent interface can directly carry out instant communication with the questioning user, thereby realizing switching from the robot to the artificial customer service and continuing the session with the questioning user.

Step S1630, pushing a notification message representing the agent switching user to the chat interface of the question user side corresponding to the session:

in order to facilitate the questioning user to know the response mechanism of the intelligent customer service system, a notification message representing the switching agent user can be constructed, and the notification message is pushed to the chat interface of the user side of the questioning user to be displayed.

The embodiment further perfects the service closed loop of the intelligent customer service system for dispatching the seat interface for the session, allows the intelligent customer service system to configure the label for the seat interface at the background, and matches the corresponding seat interface for the session by using the matching condition of the label after the hit label of the session is determined.

Referring to fig. 6, in an expanded embodiment, after the step of determining, according to the correlation matrix, a tag corresponding to a tag vector whose full-size semantic vector meets a preset correlation matching condition as a hit tag in the step S1500, the method includes the following steps:

step S1700, responding to a session management request of any agent user, and pushing a session list to the agent user, wherein the session list comprises all ongoing sessions of the intelligent customer service system and mapping relation data between the hit tags:

the embodiment further provides a mechanism for selecting the sessions suitable for processing by the agent user, so that the agent user can initiate a session management request to the intelligent customer service system, the intelligent customer service system responds to the request and pushes a session list currently existing in the system to the agent user, and the session list includes all sessions currently performed by the intelligent customer service system and mapping relationship data between the hit tags, and is displayed in a graphical user interface of a client device where the agent user is located for the agent user to refer.

Step S1800, receiving the hit tag specified by the agent user, and updating the session list so that the session list only includes the session corresponding to the specified hit tag:

when the agent user wants to process a session corresponding to a certain hit tag, one hit tag provided by the session list can be selected in the graphical user interface of the agent user, and the hit tag is submitted to the intelligent customer service system for instructing the intelligent customer service system to filter the session for the intelligent customer service system. The intelligent customer service system then filters out a new session list according to the specified hit tag, so that the session list only contains the sessions corresponding to the specified hit tag. And similarly, the updated session list is also pushed to a graphical user interface on the user side for display.

Step S1900, responding to a participation request of the agent user selecting any one session in the session list, establishing a data communication link between the agent user and the questioning user of the session, so that the agent user and the questioning user continue the session:

on the basis of the session list, the agent user can perform any session in the session list, so that a participation request is triggered and sent to the intelligent customer service system, the intelligent customer service system accordingly establishes a data communication link between the agent user and the questioning user of the session, the session is continued, and the agent user responds to the questioning of the questioning user, so that manual customer service access is realized.

It should be noted that, each step of the present embodiment may be processed concurrently with the step S1600, and the execution of the present embodiment does not depend on the execution of the step S1600.

According to the implementation of the embodiment, the function of filtering the sessions suitable for the agent user to process by self is opened for the agent user, so that the agent user can obtain the session list suitable for the agent user to process according to the hit label selected by the agent user subjectively, the agent user can reply the message of the consumer user more primarily and secondarily, the initiative can be mastered, and the response efficiency of the intelligent customer service system is improved.

Referring to fig. 7, a customer service session scheduling apparatus adapted to one of the objectives of the present application is a functional implementation of the customer service session scheduling method of the present application, and the apparatus includes: the method comprises the following steps: the system comprises a data acquisition module 1300, a semantic extraction module 1400, a tag determination module 1500, and a communication establishment module 1600, wherein the data acquisition module 1300 is configured to acquire dialog data of any session in the intelligent customer service system, where the dialog data includes a question text generated by a question user of the session and a reply text generated by the system; the semantic extraction module 1400 is configured to extract deep semantic information of a word vector corresponding to each participle from the dialogue data, and obtain a semantic vector corresponding to each word vector; the tag determining module 1500 is configured to calculate a correlation matrix between tag vectors corresponding to each tag in a preset tag library and each semantic vector, and determine, according to the correlation matrix, a tag corresponding to a tag vector whose full amount of semantic vectors meets a preset correlation matching condition as a hit tag; the communication establishing module 1600 is configured to access the session corresponding to the dialog data for which the hit tag is determined to the agent interface corresponding to the hit tag, so that the agent user corresponding to the agent interface and the questioning user of the session continue the session.

In a further embodiment, the semantic extraction module 1400 includes: the text splicing submodule is used for sequentially splicing the question text and the reply text in the dialogue data according to the created time sequence to form a dialogue text; the text specification submodule is used for carrying out preset standardized preprocessing on the conversation text to enable the conversation text to become a specification text; the text word segmentation sub-module is used for carrying out word segmentation processing on the standard text to obtain a word segmentation sequence of the standard text; the word segmentation coding sub-module is used for querying a preset word vector table, determining word vectors corresponding to all the segmentation words in the word segmentation sequence and obtaining a word vector sequence corresponding to the standard text; and the semantic extraction submodule is used for extracting deep semantic information of each word vector in the word vector sequence by using a text extraction model which is trained to be in a convergence state in advance and referring to the context information, and mapping the deep semantic information to a high-dimensional space to obtain a semantic vector corresponding to each word vector.

In a further embodiment, the tag determining module 1500 includes: the matrix construction submodule is used for calculating a correlation matrix between a label vector corresponding to each label in a preset label library and each semantic vector by adopting a preset correlation algorithm, and the correlation matrix comprises a correlation coefficient of each semantic vector mapped to each label vector; the coefficient synthesis submodule is used for calculating a synthesis correlation coefficient corresponding to all semantic vectors mapped to each label vector by applying a preset voting algorithm according to the correlation coefficient so as to represent the correlation probability of the dialogue data belonging to each label vector; and the tag hit submodule is used for screening according to preset relevant matching conditions, determining one or more tag vectors with the highest relevant probability as the tag vector(s) most relevant to the full-scale semantic vector, and taking the corresponding tag(s) as hit tags.

In a further embodiment, the communication establishing module 1600 includes: the seat determining submodule is used for inquiring the label tags corresponding to the preset seat interfaces and determining the seat interfaces of which the label tags completely contain all hit labels as the seat interfaces corresponding to the hit labels; the link establishing submodule is used for establishing a data communication link between the determined seat interface and the questioning user of the session corresponding to the dialogue data of which the hit label is determined, so that the session is continued by the seat user corresponding to the seat interface and the questioning user of the session; and the notification pushing submodule is used for pushing a notification message representing the switching agent user to a chat interface of the questioning user side corresponding to the session.

In an extended embodiment, the customer service session scheduling apparatus of the present application further includes: the initial pushing submodule is used for responding to a session management request of any agent user and pushing a session list to the agent user, wherein the session list comprises all ongoing sessions of the intelligent customer service system and mapping relation data among the hit labels; the updating and pushing submodule is used for receiving the hit label appointed by the agent user and updating the session list to ensure that the session list only contains the session corresponding to the appointed hit label; and the call switching submodule is used for responding to a participation request of any one session selected by the agent user in the session list, and establishing a data communication link between the agent user and the questioning user of the session so that the agent user and the questioning user continue the session.

In order to solve the technical problem, an embodiment of the present application further provides a computer device. As shown in fig. 8, the internal structure of the computer device is schematically illustrated. The computer device includes a processor, a computer-readable storage medium, a memory, and a network interface connected by a system bus. The computer readable storage medium of the computer device stores an operating system, a database and computer readable instructions, the database can store control information sequences, and the computer readable instructions, when executed by the processor, can make the processor implement a customer service session scheduling method. The processor of the computer device is used for providing calculation and control capability and supporting the operation of the whole computer device. The memory of the computer device may have stored therein computer readable instructions that, when executed by the processor, may cause the processor to perform the customer service session scheduling method of the present application. The network interface of the computer device is used for connecting and communicating with the terminal. Those skilled in the art will appreciate that the architecture shown in fig. 8 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

In this embodiment, the processor is configured to execute specific functions of each module and its sub-module in fig. 7, and the memory stores program codes and various data required for executing the modules or the sub-modules. The network interface is used for data transmission to and from a user terminal or a server. The memory in this embodiment stores program codes and data required for executing all modules/sub-modules in the customer service session scheduling device of the present application, and the server can call the program codes and data of the server to execute the functions of all sub-modules.

The present application further provides a storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of the customer service session scheduling method of any of the embodiments of the present application.

The present application also provides a computer program product comprising computer programs/instructions which, when executed by one or more processors, implement the steps of the method as described in any of the embodiments of the present application.

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments of the present application may be implemented by hardware related to instructions of a computer program, where the computer program may be stored in a computer-readable storage medium, and when executed, the computer program may include the processes of the embodiments of the methods as described above. The storage medium may be a computer-readable storage medium such as a magnetic disk, an optical disk, a Read-Only Memory (ROM), or a Random Access Memory (RAM).

To sum up, the intelligent degree of the intelligent customer service system can be comprehensively improved, the conversation and the seat interface can be accurately matched according to the semantic association degree of the conversation data and the preset labels of the conversation, the shunting scheduling efficiency of the intelligent customer service system is improved, waiting time is saved for questioning users, and cost reduction and efficiency improvement are realized for merchant users configuring the intelligent customer service system.

Those of skill in the art will appreciate that the various operations, methods, steps in the processes, acts, or solutions discussed in this application can be interchanged, modified, combined, or eliminated. Further, other steps, measures, or schemes in various operations, methods, or flows that have been discussed in this application can be alternated, altered, rearranged, broken down, combined, or deleted. Further, steps, measures, schemes in the prior art having various operations, methods, procedures disclosed in the present application may also be alternated, modified, rearranged, decomposed, combined, or deleted.

The foregoing is only a partial embodiment of the present application, and it should be noted that, for those skilled in the art, several modifications and decorations can be made without departing from the principle of the present application, and these modifications and decorations should also be regarded as the protection scope of the present application.

Claims

1. A customer service session scheduling method is characterized by comprising the following steps:

2. The customer service session scheduling method according to claim 1, wherein deep semantic information of a word vector corresponding to each participle is extracted from the dialogue data to obtain a semantic vector corresponding to each word vector, comprising the steps of:

carrying out preset standardized preprocessing on the conversation text to enable the conversation text to become a standard text;

3. The customer service session scheduling method according to claim 1, wherein a correlation matrix between a tag vector corresponding to each tag in a preset tag library and each semantic vector is calculated, and a tag corresponding to a tag vector whose full amount of semantic vectors meets a preset correlation matching condition is determined as a hit tag according to the correlation matrix, comprising the steps of:

4. The customer service session dispatching method according to any one of claims 1 to 3, wherein the step of calculating the correlation matrix between the tag vector corresponding to each tag in the preset tag library and each semantic vector is preceded by the steps of:

5. The customer service session scheduling method according to any one of claims 1 to 3, wherein the session corresponding to the dialogue data for which the hit tag is determined is accessed to the agent interface corresponding to the hit tag, comprising the following steps:

establishing a data communication link between the determined agent interface and the questioning user of the session corresponding to the dialogue data of which the hit tag is determined, so that the session is continued by the agent user corresponding to the agent interface and the questioning user of the session;

6. The customer service session scheduling method according to any one of claims 1 to 3, wherein after the step of determining, as hit tags, tags corresponding to tag vectors whose full semantic vectors satisfy the preset correlation matching condition according to the correlation matrix, the method comprises the steps of:

7. A customer service session scheduler, comprising:

the data acquisition module is used for acquiring dialogue data of any session in the intelligent customer service system, wherein the dialogue data comprises a question text generated by a question user of the session and a reply text generated by the system;

the semantic extraction module is used for extracting deep semantic information of word vectors corresponding to each participle from the dialogue data to obtain semantic vectors corresponding to each word vector;

the tag determining module is used for calculating a correlation matrix between a tag vector corresponding to each tag in a preset tag library and each semantic vector, and determining tags corresponding to tag vectors of which the full quantity of semantic vectors meet preset correlation matching conditions as hit tags according to the correlation matrix;

and the communication establishing module is used for accessing the session corresponding to the dialogue data with the determined hit label to the seat interface corresponding to the hit label so that the seat user corresponding to the seat interface and the questioning user of the session continue the session.

8. A computer device comprising a central processor and a memory, characterized in that the central processor is adapted to invoke execution of a computer program stored in the memory to perform the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that it stores, in the form of computer-readable instructions, a computer program implemented according to the method of any one of claims 1 to 6, which, when invoked by a computer, performs the steps comprised by the corresponding method.

10. A computer program product comprising computer program/instructions, characterized in that the computer program/instructions, when executed by a processor, implement the steps of the method as claimed in any one of claims 1 to 6.