CN110688454A

CN110688454A - Method, device, equipment and storage medium for processing consultation conversation

Info

Publication number: CN110688454A
Application number: CN201910849393.6A
Authority: CN
Inventors: 张鹭; 陈娟; 傅婧; 黄忆丁; 郭鹏程
Original assignee: OneConnect Financial Technology Co Ltd Shanghai
Current assignee: OneConnect Financial Technology Co Ltd Shanghai
Priority date: 2019-09-09
Filing date: 2019-09-09
Publication date: 2020-01-14
Also published as: WO2021047186A1

Abstract

The application relates to the field of artificial intelligence, and provides a method, a device, equipment and a storage medium for consultation dialogue processing, wherein the method comprises the following steps: acquiring a conversation consultation model, receiving target product information input by a user, and classifying the target product information through the conversation consultation model to obtain a classification result; constructing a product information base according to the classification result; when receiving the consultation information input by the user, inputting the consultation information into the conversation consultation model to obtain the consultation intention information of the user; matching second product information corresponding to the consultation intention information in the product knowledge base, converting a semantic representation form of the second product information into a conversational language, and generating and outputting a consultation return result; and counting and analyzing the historical key information and the historical consultation return result, and generating and outputting a multi-dimensional report. By adopting the scheme, the service efficiency of the enterprise service platform for product consultation can be improved.

Description

Method, device, equipment and storage medium for processing consultation conversation

Technical Field

The present application relates to the field of intelligent decision making, and in particular, to a method, an apparatus, a device, and a storage medium for processing a consultation session.

Background

In the current product sales management, a consultant generally sends a query request to a system processing center through an enterprise service platform, the system processing center analyzes the information of the query request, acquires the consultation intention information of the consultant, matches a corresponding consultation result according to the consultation intention information, returns the consultation result to the enterprise service platform, and displays the consultation result in the enterprise service platform in a text form.

Because the accuracy of the consultation intention information obtained by the current enterprise service platform for the inquiry request input by the consultant is low, the matching degree of the obtained consultation result and the consultation request is low, so that the consultant needs to input the consultation information for many times or cannot obtain the required consultation result finally. Because the prior enterprise service platform displays the consultation result in a text form, the display mode is fixed and single, so that the time for a consultant to know the consultation result displayed by the enterprise service platform is relatively long. Therefore, the service efficiency of the product consultation of the enterprise service platform is low.

Disclosure of Invention

The application provides a method, a device, equipment and a storage medium for consultation conversation processing, which can solve the problem that the service efficiency of an enterprise service platform for product consultation is low in the prior art.

In a first aspect, the present application provides a method of consultation session processing, the method comprising:

the method comprises the steps of obtaining training information, inputting the training information into a model, training the model to obtain a dialogue consultation model, analyzing and processing target product information input by a user to construct a product information base, analyzing and processing the consultation information input by the user to obtain a return result corresponding to the consultation information, wherein the training information comprises various first product information, consultation dialogue information and network knowledge information, and the dialogue consultation model comprises a first sub-model and a second sub-model;

receiving target product information input by a user, classifying the target product information through the dialogue consultation model, and obtaining a classification result, wherein the classification result comprises the attribute, the field and the production stage of a target product;

establishing a first corresponding relation and/or a second corresponding relation according to the classification result, and establishing a product information base according to the first corresponding relation or the second corresponding relation, wherein the first corresponding relation comprises the corresponding relation between target product information and target information, and the second corresponding relation comprises the corresponding relation between classification information, product characteristic information and recommendation information;

when receiving consultation information input by a user, inputting the consultation information into the first submodel, and acquiring key information from the consultation information through the first submodel;

inputting the key information into the second submodel, and analyzing the key information through the second submodel to obtain consultation intention information of the user;

matching second product information corresponding to the consultation intention information in the product knowledge base, converting a semantic representation form of the second product information into a conversational language, and generating and outputting a consultation return result;

and counting and analyzing historical key information and historical consultation return results, and generating and outputting a multi-dimensional report, wherein the historical key information comprises key information obtained from the received multi-time consultation information within preset time, and the historical consultation return results comprise consultation return results obtained from the received multi-time consultation information within preset time.

In one possible design, the obtaining training information, inputting the training information into a model, and training the model to obtain a conversational consultation model includes:

acquiring training information, inputting the training information into a neural network model, classifying the first product information and the network knowledge information according to a preset association rule, respectively acquiring first association information and second association information, and acquiring consultation target information of the consultation dialogue information;

establishing a corresponding relation among the consulting purpose information, the first related information and the second related information, outputting the first related information and the second related information when the consulting purpose information is detected by the neural network model, and inputting the consulting purpose information, the corresponding relation, the first related information and the second related information into the neural network model;

carrying out multiple groups of parameter training on the neural network model to obtain a neural network model to be trained under multiple groups of parameters;

evaluating the error of the neural network model under each set of parameters to obtain a plurality of error values;

the method comprises the steps of obtaining a minimum error value by calculating the size of a plurality of error values, taking a group of parameters corresponding to the minimum error value as parameters of a neural network model, obtaining a target neural network model, deploying the target neural network model, and obtaining a conversation consultation model, wherein the deployment refers to packaging, installing, configuring and releasing a configuration file, a user manual and a help document of the applied target neural network model.

In a possible design, the dialogue advisory model includes a regression analysis model, the method further includes, before the establishing a first corresponding relationship and/or a second corresponding relationship according to the classification result and the building a product information base according to the first corresponding relationship or the second corresponding relationship:

inputting analysis data to the dialogue and consultation model, wherein the analysis data comprises information and purchase information of product users, information of product sales conditions, consumption levels and consumption directions of various places and purchase conditions of various places in the product field;

performing regression analysis on the analysis data through the regression analysis model to obtain recommendation information, wherein the recommendation information comprises development tendency of products, developable regions, marketable people and product information outside the analysis products in the field to which the products belong;

the establishing a first corresponding relationship and/or a second corresponding relationship according to the classification result, and the establishing a product information base according to the first corresponding relationship or the second corresponding relationship comprise:

classifying the target product information according to the product function, the product type and the category of the product structure to obtain classification information;

acquiring first characteristic information of the classification information;

and establishing a second corresponding relation among the classification information, the first characteristic information and the recommendation information to obtain a product information base.

In one possible design, the conversational consultation model includes a capsule network model, the inputting the key information into the second submodel, the analyzing the key information by the second submodel to obtain consultation intention information of the user includes:

at least one of the following text data processing modes is carried out on the key information to obtain the target consultation information: the method comprises the following steps of field analysis, text error correction, text completion, reference resolution, word decomposition, part of speech tagging, entity recognition and text feature extraction;

performing intention classification on the target consultation information through the capsule network model to acquire intention classification information;

and acquiring attribute information in the intention classification information, and performing reasoning and context decision on the attribute information to acquire consultation intention information of the user.

In one possible design, before the text data processing is performed on the key information, the method further includes:

acquiring training data, wherein the training data comprises statement query language and query consultation language;

performing attribute analysis on the training data, and marking the training data subjected to the attribute analysis to obtain words with marked attributes, wherein the attribute analysis comprises proper noun attribute analysis, gender attribute analysis, single-complex attribute analysis, distance attribute analysis and abbreviated matching attribute analysis;

acquiring second feature information of the training data, and selecting one feature from the second feature information as a splitting standard of a node, wherein a calculation formula for acquiring the splitting standard is as follows:

wherein info (D) is entropy, G (A) is information gain rate, D is the training data, m is the number of words of the labeled attributes, p_iA word is the mark attribute corresponding to a selected feature in the second feature information, a is one of the mark attributes, and v is the output number and the output division number corresponding to the attribute a test;

taking the words with the marked attributes as root nodes, and constructing a decision tree according to a preset classification standard and the splitting standard;

the text data processing of the key information comprises:

performing attribute analysis on the key information, marking the key information subjected to attribute analysis, and acquiring a target word with marked attributes;

and performing data analysis and processing on the target words through the decision tree.

In one possible design, after the matching of the second product information corresponding to the consultation intention information in the product knowledge base, the method further includes:

obtaining a matching degree, wherein the matching degree is the corresponding degree of the product information in the product knowledge base and the consultation intention information;

detecting whether the matching degree is smaller than a first preset threshold value or not;

when the detection result is yes, analyzing the consultation information to obtain information to be solved, wherein the information to be solved comprises the field, the name, the technical problem and the problem type of the product;

matching consultation answering personnel according to the information to be solved, and sending the consultation information and the prompt information limited by answering time to an information acquisition tool so that the information acquisition tool generates a solution result according to the consultation information and the prompt information;

and acquiring the solution result received by the information acquisition tool, inputting the solution information and the solution result into the dialogue consultation model, training the dialogue consultation model, and acquiring the updated dialogue consultation model.

acquiring a plurality of matching values, wherein the matching values are corresponding degree values of the product information in the product knowledge base and the consultation intention information;

calculating a plurality of matching values, and comparing the sizes of the matching values to obtain a maximum matching value;

detecting whether the maximum matching value is smaller than a second preset threshold value or not;

and if so, outputting target recommendation information, wherein the target recommendation information comprises product information corresponding to the maximum matching value, recommendation information matched with the product information and product information related to the product information.

In a second aspect, the present application provides an apparatus for consultation session processing having functions of implementing a method corresponding to the consultation session processing provided in the first aspect described above. The functions can be realized by hardware, and the functions can also be realized by executing corresponding software by hardware. The hardware or software includes one or more modules corresponding to the above functions, which may be software and/or hardware.

In one possible design, the apparatus includes:

the input and output module is used for acquiring training information, receiving target product information input by a user and receiving consultation information input by the user;

the processing module is used for inputting the training information acquired by the input and output module into a model, training the model to acquire a dialogue consultation model, analyzing and processing the target product information input by the user to construct a product information base, and analyzing and processing the consultation information input by the user to acquire a return result corresponding to the consultation information; classifying the target product information received by the input and output module through the dialogue consultation model to obtain a classification result; establishing a first corresponding relation and/or a second corresponding relation according to the classification result, and establishing a product information base according to the first corresponding relation or the second corresponding relation; inputting the consultation information received by the input and output module into the first submodel, and acquiring key information from the consultation information through the first submodel; inputting the key information into the second submodel, and analyzing the key information through the second submodel to obtain consultation intention information of the user; matching second product information corresponding to the consultation intention information in the product knowledge base, converting a semantic representation form of the second product information into a conversational terminology, generating a consultation return result, and inputting the consultation return result to a display module through the input and output module; counting and analyzing historical key information and historical consultation return results, generating a multi-dimensional report and inputting the multi-dimensional report to the display module through the input and output module;

and the display module is used for receiving and displaying the consultation return result and the multi-dimensional report form from the input and output module.

In one possible design, the processing module is specifically configured to:

In a possible design, before executing the dialogue consultation model including a regression analysis model, the processing module is further configured to, before establishing the first corresponding relationship and/or the second corresponding relationship according to the classification result, and constructing the product information base according to the first corresponding relationship or the second corresponding relationship:

acquiring first characteristic information of the classification information;

In one possible design, the processing module is specifically configured to:

In one possible design, before performing the text data processing on the key information, the processing module is further configured to:

the text data processing of the key information comprises:

In one possible design, the processing module, after performing the matching of the second product information corresponding to the consultation intention information in the product knowledge base, is further configured to:

A further aspect of the application provides a computer device comprising at least one connected processor, memory, display and input-output unit, wherein the memory is configured to store program code and the processor is configured to call the program code in the memory to perform the method of the first aspect.

A further aspect of the present application provides a computer storage medium comprising instructions which, when run on a computer, cause the computer to perform the method of the first aspect described above.

Compared with the prior art, in the scheme provided by the application, the target product information input by the user is received, and the dialogue consultation model is used for classifying the target product information to obtain a classification result; constructing a product information base according to the classification result; when receiving the consultation information input by the user, inputting the consultation information into the conversation consultation model to obtain the consultation intention information of the user; matching second product information corresponding to the consultation intention information in the product knowledge base, converting a semantic representation form of the second product information into a conversational language, and generating and outputting a consultation return result; and counting and analyzing the historical key information and the historical consultation return result, and generating and outputting a multi-dimensional report. The product information is sorted and classified to construct a product information base so as to quickly and accurately acquire the product information corresponding to the consultation information, and the product information related to the product information corresponding to the consultation information is acquired in the product information base, so that the consultation return result is displayed in multiple aspects and multiple angles, and the accuracy and the versatility of the consultation return result are improved; the consultation return result is displayed in a language and a multi-dimensional report form, so that the user can understand the consultation return result from multiple angles, and the acquisition time of the user for the content of the consultation return result is shortened. By improving the accuracy and the speed of acquiring the consultation return result and shortening the time for acquiring the content of the consultation return result by the user, the consultation times and the consultation time of the user on the enterprise service platform are reduced, and the service efficiency of the enterprise service platform for product consultation is improved.

Drawings

FIG. 1 is a schematic flow chart of a method for consultation session processing in an embodiment of the present application;

FIG. 2 is an explanatory diagram of a consultation return result in the embodiment of the present application;

FIG. 3 is an explanatory diagram of target product information in the embodiment of the present application;

FIG. 4 is a schematic structural diagram of a first sub-model in the embodiment of the present application;

FIG. 5 is a schematic structural diagram of a first sub-model in an embodiment of the present application;

FIG. 6 is a schematic structural diagram of an apparatus for consultation session processing in the embodiment of the present application;

fig. 7 is a schematic structural diagram of a computer device according to an embodiment of the present application.

The implementation, functional features and advantages of the objectives of the present application will be further explained with reference to the accompanying drawings.

Detailed Description

It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application. The terms "first," "second," and the like in the description and in the claims of the present application and in the above-described drawings are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It will be appreciated that the data so used may be interchanged under appropriate circumstances such that the embodiments described herein may be practiced otherwise than as specifically illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or modules is not necessarily limited to those steps or modules explicitly listed, but may include other steps or modules not explicitly listed or inherent to such process, method, article, or apparatus, and such that a division of modules presented in this application is merely a logical division that may be implemented in an actual application in a different manner, such that multiple modules may be combined or integrated into another system, or some features may be omitted, or may not be implemented.

The application provides a method, a device, equipment and a storage medium for consultation dialogue processing, which can be used for consultation management of an enterprise intelligent platform on products.

In order to solve the technical problems, the application mainly provides the following technical scheme:

classifying the target product information through the dialogue consultation model by receiving the target product information input by a user to obtain a classification result; constructing a product information base according to the classification result; when receiving the consultation information input by the user, inputting the consultation information into the conversation consultation model to obtain the consultation intention information of the user; matching second product information corresponding to the consultation intention information in the product knowledge base, converting a semantic representation form of the second product information into a conversational language, and generating and outputting a consultation return result; and counting and analyzing the historical key information and the historical consultation return result, and generating and outputting a multi-dimensional report. The product information is sorted and classified to construct a product information base so as to quickly and accurately acquire the product information corresponding to the consultation information, and the product information related to the product information corresponding to the consultation information is acquired in the product information base, so that the consultation return result is displayed in multiple aspects and multiple angles, and the accuracy and the versatility of the consultation return result are improved; the consultation return result is displayed in a language and a multi-dimensional report form, so that the user can understand the consultation return result from multiple angles, and the acquisition time of the user for the content of the consultation return result is shortened. By improving the accuracy and the speed of acquiring the consultation return result and shortening the time for acquiring the content of the consultation return result by the user, the consultation times and the consultation time of the user on the enterprise service platform are reduced, and the service efficiency of the enterprise service platform for product consultation is improved.

Referring to fig. 1, a method for providing a consultation session process according to the present application is illustrated as follows, the method including:

101. and acquiring training information, inputting the training information into the model, and training the model to acquire the dialogue consultation model.

The system comprises a dialogue consultation model, a product information database and a database, wherein the dialogue consultation model is used for analyzing and processing target product information input by a user to construct the product information database and analyzing and processing consultation information input by the user to obtain a return result corresponding to the consultation information; the training information comprises various first product information, consultation dialogue information and network knowledge information; the conversational consultation model includes a first submodel and a second submodel.

The network knowledge information includes: first other product information other than the first product information, which is the same as and/or similar to a field to which a product of the first product information belongs, and second other product information other than the first product information, which is the same as and/or similar to a type to which a product of the product information belongs.

The dialogue consultation model is used for analyzing and processing target product information input by a user to construct a product information base, and analyzing and processing consultation information input by the user to obtain a consultation return result (or a related return result) corresponding to the consultation information.

In some embodiments, the consultation return result includes information on the product being consulted and information on recommended products that are the same as or similar to the domain, type, and specific performance to which the product being consulted belongs, as shown in fig. 2. The contents of fig. 2 are only referred to by way of example, and the content accuracy and actual operation are not considered.

102. And receiving target product information input by a user, classifying the target product information through a dialogue consultation model, and acquiring a classification result.

The classification result comprises the attribute, the field and the production stage of the target product.

The target product information comprises various information of the product, responsible personnel of the product and workflow node information of the responsible personnel, and the workflow node information comprises project making and decision making of the product, deployment of various works in the early stage of the product, deployment of various works in the middle stage of the product and deployment of various works in the later stage of the product. For example: when a user inputs a good welfare APP product in a terminal device, the dialogue and consultation model outputs product information such as functions, structures and performances related to the good welfare APP product, personal information of a product research and development team and contents responsible for the personal information, as shown in fig. 3. The contents of fig. 3 are only given as examples, and the content accuracy and actual operation are not considered.

103. And establishing a first corresponding relation and/or a second corresponding relation according to the classification result, and establishing a product information base according to the first corresponding relation or the second corresponding relation.

The first corresponding relation comprises the corresponding relation between the target product information and the target information, and the second corresponding relation comprises the corresponding relation between the classification information, the product characteristic information and the recommendation information.

In some embodiments, the conversational consulting model includes a knowledge base system created from the collected mass network data, which may be used to build a product information base on the basis of the knowledge base system.

By constructing the product information base, the target information corresponding to the target product information can be rapidly and accurately acquired according to the target product information.

104. When receiving the consultation information input by the user, inputting the consultation information into the first submodel, and acquiring key information from the consultation information through the first submodel.

The first sub-model comprises an image recognition sub-model, a text recognition sub-model and an audio-video recognition sub-model, and the first sub-model further comprises a classifier. The image recognition sub-model, the text recognition sub-model and the audio and video recognition sub-model are connected in parallel, and the image recognition sub-model, the text recognition sub-model and the audio and video recognition sub-model are respectively connected with the classifier in series.

For example: the consultation information input by the user in the APP is a product picture, the consultation dialogue model receives the product picture and inputs the product picture into the classifier, the classifier identifies the product picture, the product picture is input into the image identification sub-model, the image identification sub-model identifies and acquires the shape of a product in the product picture, a corresponding product image is matched in the product image database according to the acquired shape of the product, a product name corresponding to the product image is acquired, and the product name is used as key information of the consultation information, as shown in fig. 4.

In some embodiments, the first sub-model further comprises a plurality of filters, filter 1 in the image recognition sub-model, filter 2 in the text recognition sub-model and filter 3 in the audiovisual recognition sub-model. The image recognition sub-model, the text recognition sub-model and the audio and video recognition sub-model are connected in parallel, and the image recognition sub-model, the text recognition sub-model and the audio and video recognition sub-model are respectively connected with a filter in series.

For example: the consultation information input by a user in the APP is a product picture, the consultation dialogue model receives the product picture and inputs the product picture into an image recognition sub-model, a text recognition sub-model and an audio-video recognition sub-model, respective filters in the image recognition sub-model, the text recognition sub-model and the audio-video recognition sub-model recognize the product picture, if the filter in the image recognition sub-model recognizes that the input information is a picture, the image recognition sub-model continues working to recognize and acquire the shape of a product in the product picture, a corresponding product image is matched in a product image database according to the acquired shape of the product, a product name corresponding to the product image is acquired, and the product name is used as key information of the consultation information; if the filter in the text recognition submodel recognizes that the input information is non-text information, the text recognition submodel does not output; the audio and video recognition submodel has the same principle. As shown in fig. 5.

According to the method and the device, product information can be consulted in a mode of carrying out conversation consultation, text input and picture input with the robot through the online video, and therefore, the method and the device can realize multi-way input to improve user experience.

105. And inputting the key information into a second submodel, and analyzing the key information through the second submodel to obtain the consultation intention information of the user.

The second sub-model is used for analyzing and processing the key information to acquire the consultation intention information of the user. The second sub-model analyzes the received key information and obtains the consultation intention information of the user, and the obtained consultation intention information of the user can comprise various contents.

For example: the second submodel receives key information of 'memory card, use', analyzes and obtains the information of the user's consultation intention possibly corresponding to the key information of' memory card, use ', what product the memory card can match with for use' and 'main matters of memory card use and related problems encountered in the use process and solutions thereof'.

106. And matching second product information corresponding to the consultation intention information in the product knowledge base, converting the semantic representation form of the second product information into a conversational terminology, and generating and outputting a consultation return result.

By converting the semantic representation form of the second product information into the conversational language, the time for the consultant to conceive the sales expression form of the language of the second product information is reduced, thereby facilitating quick use of the second product information by the consultant.

107. And counting and analyzing the historical key information and the historical consultation return result, and generating and outputting a multi-dimensional report.

The historical key information comprises key information obtained from the received multi-time consultation information within the preset time, and the historical consultation return result comprises a consultation return result obtained from the received multi-time consultation information within the preset time.

The multi-dimensional report at least comprises the following three items: the number of users for consultation, the type of consultation content, the number of times of consultation, repeated consultation information, consultation information processed manually in the background, and consultation information processed and unprocessed manually in the background.

Compared with the existing mechanism, in the embodiment of the application, the target product information input by the user is received, and the target product information is classified through the dialogue consultation model to obtain a classification result; constructing a product information base according to the classification result; when receiving the consultation information input by the user, inputting the consultation information into a conversation consultation model to obtain the consultation intention information of the user; matching second product information corresponding to the consultation intention information in a product knowledge base, converting the semantic representation form of the second product information into a conversational terminology, and generating and outputting a consultation return result; and counting and analyzing the historical key information and the historical consultation return result, and generating and outputting a multi-dimensional report. The product information is sorted and classified to construct the product information base so as to quickly and accurately acquire the product information corresponding to the consultation information, and the product information related to the product information corresponding to the consultation information is acquired in the product information base, so that the consultation return result is displayed in multiple aspects and multiple angles, and the accuracy and the versatility of the consultation return result are improved; the consultation return result is displayed in a language and a multi-dimensional report form, so that the user can understand the consultation return result from multiple angles, and the acquisition time of the user for the content of the consultation return result is shortened. By improving the accuracy and the speed of acquiring the consultation return result and shortening the time for acquiring the content of the consultation return result by the user, the consultation times and the consultation time of the user on the enterprise service platform are reduced, and the service efficiency of the enterprise service platform for product consultation is improved.

Optionally, in some embodiments of the present application, the obtaining training information, inputting the training information into the model, and training the model to obtain the dialogue consultation model includes:

acquiring training information, inputting the training information into a neural network model, classifying first product information and network knowledge information according to a preset association rule, acquiring first association information and second association information respectively, and acquiring consultation target information of consultation dialogue information;

establishing a corresponding relation of the consultation target information, the first associated information and the second associated information, outputting the first associated information and the second associated information when the consultation target information is detected by the neural network model, and inputting the consultation target information, the corresponding relation, the first associated information and the second associated information into the neural network model;

training a plurality of groups of parameters of the neural network model to obtain the neural network model to be trained under the plurality of groups of parameters;

the method comprises the steps of obtaining a minimum error value by calculating the size of a plurality of error values, taking a group of parameters corresponding to the minimum error value as parameters of a neural network model, obtaining a target neural network model, deploying the target neural network model, and obtaining a conversation consultation model, wherein the deploying refers to packaging, installing, configuring and releasing a configuration file, a user manual and a help document of the applied target neural network model.

The neural network model is composed of input nodes, output nodes and a node layer. The parameters include at least one of a size of a convolution kernel, a learning rate η, a batch parameter, a number of neural network layers, an activation function, an optimizer, a batch size, and a number of training hyper-parameters epochs.

The accuracy of the neural network model is improved by carrying out error evaluation on the neural network model to be trained under multiple groups of parameters. And deploying the neural network model subjected to error evaluation to obtain a dialogue consultation model with relatively complete function and relatively stable performance.

Optionally, in some embodiments of the application, the above-mentioned dialogue consultation model includes a regression analysis model, the first corresponding relationship and/or the second corresponding relationship are/is established according to the classification result, and before the product information base is constructed according to the first corresponding relationship or the second corresponding relationship, the method includes:

inputting analysis data to the dialogue consultation model, wherein the analysis data comprises information and purchase information of product users, information of product sales conditions, consumption levels and consumption directions of various places and purchase conditions of various places in the product field;

performing regression analysis on the analysis data through a regression analysis model to obtain recommendation information, wherein the recommendation information comprises development tendency of products, developable regions, marketable people and product information except for the analysis products in the field to which the products belong;

establishing a first corresponding relation and/or a second corresponding relation according to the classification result, and establishing a product information base according to the first corresponding relation or the second corresponding relation, wherein the steps comprise:

acquiring first characteristic information of the classification information;

The recommendation information of multiple types of contents is obtained by carrying out regression analysis on the analysis data comprising the information and purchase information of product users, the condition information of product sales, the consumption level and consumption direction of each place and the purchase condition of each place in the product field, so that the multifacetability of the contents of the consultation return result of the conversation consultation module is enhanced, and the practicability and the service efficiency of the enterprise service platform corresponding to the conversation consultation module are improved.

The classification information is obtained by classifying the target product information, and then the corresponding relation of the classification information is established to establish a product information base, so that the information classification of the product information base is refined, and the efficiency and the accuracy of obtaining the information in the product information base are improved.

Optionally, in some embodiments of the present application, the above-mentioned dialogue consultation model includes a capsule network model, the key information is input into the second sub-model, and the key information is analyzed by the second sub-model to obtain the consultation intention information of the user, including:

and at least performing one of the following text data processing modes on the key information to acquire the target consultation information: the method comprises the following steps of field analysis, text error correction, text completion, reference resolution, word decomposition, part of speech tagging, entity recognition and text feature extraction;

intention classification is carried out on the target consultation information through the capsule network model so as to facilitate intention classification information;

and acquiring attribute information in the intention classification information, and performing reasoning and context decision on the attribute information to acquire the consultation intention information of the user.

The capsule network model has the characteristics of less required training data and better classification identification degree, and the acquired information has coherence through context decision so as to be more accurate, so that the capsule network model is used for carrying out intention classification on the target consultation information and carrying out reasoning and context decision according to intention classification information so as to improve the accuracy of acquiring the consultation intention information of the user.

Optionally, in some embodiments of the application, before the text data processing is performed on the key information, the method further includes:

acquiring second feature information of the training data, and selecting one feature from the second feature information as a splitting standard of the node, wherein a calculation formula for acquiring the splitting standard is as follows:

wherein, info (D) is entropy, G (A) is information gain rate, D is training data, m is number of words with marked attributes, p_iThe word is a mark attribute corresponding to a selected feature in the second feature information, A is one of the mark attributes, and v is the output number of the test corresponding to the attribute A and the division number of the output;

taking the words marked with the attributes as root nodes, and constructing a decision tree according to a preset classification standard and a splitting standard;

the text data processing is carried out on the key information, and the method further comprises the following steps:

performing attribute analysis on the key information, marking the key information subjected to the attribute analysis, and acquiring a target word with marked attributes;

and carrying out data analysis and processing on the target words through the decision tree.

The attribute is selected through the information gain rate, the defect that the attribute with a large number of values is selected in the biased direction when the attribute is selected through the information gain is overcome, pruning is carried out in the tree construction process, discretization processing of continuous attributes is completed, and incomplete data is processed.

The words with the marked attributes are used as root nodes, and the decision tree is constructed according to the preset classification standard and the splitting standard, so that the information of each node of the constructed decision tree corresponding to the target words is multi-angle and comprehensive, and the accuracy of obtaining the corresponding user consultation intention information according to the target words is improved.

Optionally, in some embodiments of the application, after the second product information corresponding to the consultation intention information is matched in the product knowledge base, the method further includes:

matching the consultation and answering personnel according to the information to be solved, and sending the consultation information and the prompt information limited by answering time to the information acquisition tool so that the information acquisition tool can generate a solution result according to the consultation information and the prompt information;

and acquiring a solution result received by the information acquisition tool, inputting the solution information and the solution result into the dialogue consultation model, training the dialogue consultation model, and acquiring the updated dialogue consultation model.

And comparing the matching degree of the product information and the consultation intention information in the product knowledge base with a first preset threshold value, returning the product information and the consultation intention information to relevant workers when the correlation of the information obtained by the product information in the product knowledge base is not high so as to obtain more accurate information, training the solution result returned by the obtained workers, and automatically obtaining and outputting the corresponding consultation return result by the conversational consultation model when relevant information to be solved is identified. Through the operation, the accuracy of the consultation return result and the practicability of the conversation consultation model are improved, and the service efficiency of the enterprise service platform corresponding to the conversation consultation model is improved.

For example: the consultation information is financial APP unresponsive, the consultation intention information is a solution corresponding to the financial APP unresponsive to analysis processing of the information, and the first preset threshold is not less than 75%. The matching degree of the product information acquired information of the product information of the consultation intention information of the solution corresponding to the analysis and processing unresponsiveness of the financing APP to the information in the product knowledge base is 40%, 40% is less than 75%, the consultation information is analyzed to acquire the information to be solved of how the analysis and processing unresponsiveness of the financing APP to the information reflects and the information to be solved of how the service management platform reflects and the financing APP-fault problem-background operation unresponsiveness-how to operate, the information to be solved and the prompt information of answering time limit are sent to the responsible person related to the technical department in the form of mail, after the solution result returned by the responsible person related to the technical department is received, the solution information and the solution result are input to a conversation consultation model, and the conversation consultation model is trained to acquire the updated conversation model. And when the conversation consultation model detects the same or similar information to be solved again, automatically acquiring and outputting a solving result.

acquiring a plurality of matching values, wherein the matching values are corresponding degree values of product information and consultation intention information in a product knowledge base;

By comparing the maximum matching value of a plurality of product information related to the consultation intention information, which is acquired in the product knowledge base according to the consultation intention information, with the size of the second preset threshold value, the target recommendation result is correspondingly output according to the judgment result, the accuracy or the relevance of the consultation return result is improved, so that a consultant can conveniently acquire the information of the consultation return result and reduce the consultation conversation times of the consultation service platform, and further, the service efficiency of the enterprise service platform corresponding to the conversation consultation model is improved.

Optionally, in some embodiments of the present application, the evaluating the error of the neural network model under each set of parameters to obtain a plurality of error values includes:

respectively obtaining a test value and a theoretical value of the neural network model under each group of parameters;

calculating a test value and a theoretical value, and acquiring an error of the neural network model under each group of parameters to acquire a plurality of error values, wherein a calculation formula for acquiring the plurality of error values is as follows:

wherein, H (p, q) is a cross entropy cost function, p (x) is a theoretical value, q (x) is a test value, and N is the number of samples.

The test value refers to the actual accuracy of the returned result corresponding to the advisory dialogue information output in the test set, and the theoretical value refers to the expected accuracy of the returned result corresponding to the advisory dialogue information output in the test set.

And calculating an error value by calculating a cross entropy cost function so as to measure the prediction quality of the neural network model.

marking the technical field, the product type, the product attribute and the product description type of various first product information, and marking the technical field, the product type, the product attribute and the product description type of network knowledge information;

extracting first characteristic information of the first product information and second characteristic information of the network knowledge information according to the marked content, and respectively converting semantic representation forms of the first characteristic information and the second characteristic information into semantic representation forms according to product sales jargon so as to obtain the first characteristic jargon information and the second characteristic jargon information;

acquiring keywords of the consultation dialogue information, and establishing a corresponding relation among the keywords, the first characteristic dialogue information and the second characteristic dialogue information, wherein the corresponding relation is used for acquiring the first characteristic dialogue information and the second characteristic dialogue information when the keywords are detected;

and enabling the dialogue consultation model to meet a preset rule condition through the keywords, the first characteristic language information, the second characteristic language information and the corresponding relation so as to obtain a corresponding consultation return result, wherein the preset rule condition comprises the steps of sorting and analyzing target product information input by a user so as to construct a product information base and obtaining corresponding product information by analyzing the consultation information input by the user.

Optionally, in some embodiments of the application, the dialog consultation model includes a created information base, the establishing a first corresponding relationship and/or a second corresponding relationship according to the classification result, and the constructing a product information base according to the first corresponding relationship or the second corresponding relationship includes:

acquiring the product type, the product field and the product attribute of the target product information through a dialogue consultation model;

acquiring associated product information corresponding to the product type and the field from an information base, and taking the associated product information as target information;

establishing a first corresponding relation between target product information and target information through a dialogue consultation model, classifying the target product information according to product attributes, acquiring classification information, and constructing a product information base according to the classification information and the first corresponding relation.

For example: and if the target product information is trusts and financing information, analyzing the trusts and financing information through a dialogue consultation model, wherein the obtained product type is an equity finance service type, the product field is a finance field, the product attributes comprise characteristics corresponding to definitions, income types and trusts and financing types, and the trusts and financing information is classified according to the income types to obtain classification information. The information of Chinese industrial and commercial banks with the equity finance service and the information of the Chinese industrial and commercial banks with the equity finance service products with considerable competitiveness are obtained from the information base, the equity finance service products of the Chinese industrial and commercial banks are classified into fixed income types, warranty types and non-warranty floating income types, and various types of characteristic information (such as the content of the products and the suitable types of people) are obtained. And establishing a first corresponding relation between the trust financing information and the information of the China industrial and commercial bank on the public financing service products. When the dialogue consultation model receives the input 'trust and financing', the dialogue consultation model outputs various information of trust and financing, and outputs various information of China industrial and commercial banks for the financial service products in a recommendation information mode.

Technical features mentioned in any of the embodiments or implementations of fig. 1 to 5 are also applicable to the embodiments corresponding to fig. 6 and 7 in the present application, and similar parts are not repeated.

A method of the consultation session process in the present application is explained above, and the apparatus 60 for performing the method of the consultation session process is described below.

Fig. 6 is a schematic structural diagram of an apparatus 60 for consultation session processing, which can be applied to consultation management of products by an enterprise intelligent platform. The apparatus 60 in the embodiment of the present application is capable of implementing the steps of the method of processing the consultation session corresponding to the method implemented in any one of the embodiments or implementations of fig. 1 to 5. The functions implemented by the apparatus 60 may be implemented by hardware, or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions, which may be software and/or hardware. The apparatus 60 may include an input/output module 601, a processing module 602, and a display module 603, where the functions of the input/output module 601, the processing module 602, and the display module 603 may be implemented by referring to operations performed in any one of the embodiments or implementations of fig. 1 to 5, which are not described herein again. The processing module 602 may be configured to control input and output operations of the input and output module 601, and the display module 603 may be configured to display the processing operations of the processing module 602.

In some embodiments, the input/output module 601 may be configured to obtain training information, receive target product information input by a user, and receive consultation information input by the user;

the processing module 602 is configured to input the training information acquired by the input/output module 601 into the model, train the model, and acquire a dialogue consultation model, where the dialogue consultation model is configured to analyze and process target product information input by a user to construct a product information base, and analyze and process consultation information input by the user to acquire a return result corresponding to the consultation information; classifying the target product information received by the input and output module 601 through a dialogue consultation model to obtain a classification result; establishing a first corresponding relation and/or a second corresponding relation according to the classification result, and establishing a product information base according to the first corresponding relation or the second corresponding relation; inputting the consultation information received by the input and output module 601 into a first submodel, and acquiring key information from the consultation information through the first submodel; inputting the key information into a second submodel, and analyzing the key information through the second submodel to obtain the consultation intention information of the user; matching second product information corresponding to the consultation intention information in the product knowledge base, converting the semantic representation form of the second product information into a conversational terminology, generating a consultation return result, and inputting the consultation return result to the display module 603 through the input and output module 601; counting and analyzing historical key information and historical consultation return results, generating a multi-dimensional report and inputting the multi-dimensional report to a display module 603 through an input and output module 601;

and the display module 603 is configured to receive and display the consultation return result and the multi-dimensional report from the input and output module 601.

Wherein the training information includes a plurality of first product information, consultation session information, and network knowledge information. The conversational consultation model includes a first submodel and a second submodel.

In some embodiments, the consultation return result includes information on the product being consulted and information on recommended products that are the same as or similar to the domain, type, and specific performance to which the product being consulted belongs.

The target product information comprises various information of the product, responsible personnel of the product and workflow node information of the responsible personnel, and the workflow node information comprises project making and decision making of the product, deployment of various works in the early stage of the product, deployment of various works in the middle stage of the product and deployment of various works in the later stage of the product. For example: when a user inputs 'good welfare APP product' into the terminal equipment, the dialogue consultation model outputs product information such as functions, structures and performances related to the good welfare APP product, personal information of a product research and development team and content responsible for the personal information.

In the embodiment of the application, the processing module 602 classifies the target product information input by the user through the input/output module 601 and the dialogue consultation model to obtain a classification result; constructing a product information base according to the classification result; when receiving the consultation information input by the user and acquired by the input and output module 601, inputting the consultation information into the dialogue consultation model to acquire the consultation intention information of the user; matching second product information corresponding to the consultation intention information in the product knowledge base, converting the semantic representation form of the second product information into a conversational terminology, generating and outputting a consultation return result through the display module 603; and counting and analyzing the historical key information and the historical consultation return result, and generating and outputting a multi-dimensional report through the display module 603. The product information is sorted and classified to construct the product information base so as to quickly and accurately acquire the product information corresponding to the consultation information, and the product information related to the product information corresponding to the consultation information is acquired in the product information base, so that consultation return results are displayed in multiple aspects and multiple angles, and the accuracy and the versatility of the consultation return results are improved; the consultation return result is displayed in a language and a multi-dimensional report form, so that the user can understand the consultation return result from multiple angles, and the acquisition time of the user for the content of the consultation return result is shortened. By improving the accuracy and the speed of acquiring the consultation return result and shortening the time for acquiring the content of the consultation return result by the user, the consultation times and the consultation time of the user on the enterprise service platform are reduced, and the service efficiency of the enterprise service platform for product consultation is improved.

Optionally, in some embodiments of the present application, technical features mentioned in any embodiment or implementation of the method for processing an advisory conversation also apply to the apparatus 60 for performing the method for processing an advisory conversation in the present application, and similar parts are not described again in the following.

The apparatus 60 in the embodiment of the present application is described above from the perspective of the modular functional entity, and the following describes a computer apparatus from the perspective of hardware, as shown in fig. 7, which includes: a processor, a memory, a display, an input-output unit (which may also be a transceiver, not identified in fig. 7), and a computer program stored in the memory and executable on the processor. For example, the computer program may be a program corresponding to the method for processing the consultation session in any one of the embodiments or implementations of fig. 1 to 5. For example, when the computer device implements the functions of the device 60 shown in fig. 6, the processor executes the computer program to implement the steps of the method for processing the consultation session performed by the device 60 in the embodiment corresponding to fig. 6; alternatively, the processor implements the functions of the modules in the apparatus 60 according to the embodiment corresponding to fig. 6 when executing the computer program. For another example, the computer program may be a program corresponding to the method for processing the consultation session in any one of the embodiments or implementation manners of fig. 1 to 5.

The Processor may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic, discrete hardware components, etc. The general purpose processor may be a microprocessor or the processor may be any conventional processor or the like which is the control center for the computer device and which connects the various parts of the overall computer device using various interfaces and lines.

The memory may be used to store the computer programs and/or modules, and the processor may implement various functions of the computer device by running or executing the computer programs and/or modules stored in the memory and invoking data stored in the memory. The memory may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area may store data (such as audio data, video data, etc.) created according to the use of the cellular phone, etc. In addition, the memory may include high speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), at least one magnetic disk storage device, a Flash memory device, or other volatile solid state storage device.

The input-output unit may also be replaced by a receiver and a transmitter, which may be the same or different physical entities. When they are the same physical entity, they may be collectively referred to as an input-output unit. The input-output unit may be a transceiver.

The memory may be integrated in the processor or may be provided separately from the processor.

Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solutions of the present application may be embodied in the form of a software product, which is stored in a storage medium (e.g., ROM/RAM), and includes several instructions for enabling a terminal (e.g., a mobile phone, a computer, a server, or a network device) to execute the method according to the embodiments of the present application.

The embodiments of the present application have been described above with reference to the drawings, but the present application is not limited to the above-mentioned embodiments, which are only illustrative and not restrictive, and those skilled in the art can make many changes and modifications without departing from the spirit and scope of the present application and the protection scope of the claims, and all changes and modifications that come within the meaning and range of equivalency of the claims are to be embraced within their scope.

Claims

1. A method of consultation session processing, the method comprising:

receiving target product information input by a user, and classifying the target product information through the dialogue consultation model to obtain a classification result, wherein the classification result comprises the attribute, the field and the production stage of a target product;

establishing a first corresponding relation and/or a second corresponding relation according to the classification result, and establishing a product information base according to the first corresponding relation and/or the second corresponding relation, wherein the first corresponding relation comprises the corresponding relation between target product information and target information, and the second corresponding relation comprises the corresponding relation between classification information, product characteristic information and recommendation information;

2. The method of claim 1, wherein the obtaining training information, inputting the training information into a model, and training the model to obtain a conversational consultation model comprises:

3. The method of claim 1, wherein the conversational consultation model includes a regression analysis model, and wherein before building a product information base from the first correspondence or the second correspondence, the method further comprises:

acquiring first characteristic information of the classification information;

4. The method of claim 1, wherein the conversational consultation model includes a capsule network model, the inputting the key information into the second submodel, the analyzing the key information by the second submodel to obtain the user's consultation intention information includes:

5. The method of claim 4, wherein prior to the text data processing the key information, the method further comprises:

wherein info (D) is entropy, G (A) is information gain rate, D is the training data, m is the number of words of the labeled attributes, p_iIs a word of the tag attribute corresponding to a selected one of the second feature information, a is one of the tag attributes, v is a corresponding oneTesting the output number and the output division number in the attribute A;

the text data processing of the key information comprises:

6. The method of claim 1, wherein after matching second product information corresponding to the consultation intention information in the product knowledge base, the method further comprises:

7. The method of claim 1, wherein after matching second product information corresponding to the consultation intention information in the product knowledge base, the method further comprises:

8. An apparatus for consultation session processing, the apparatus comprising:

9. A computer device, characterized in that the computer device comprises:

at least one processor, a memory, a display, and an input-output unit;

wherein the memory is configured to store program code and the processor is configured to invoke the program code stored in the memory to perform the method of any of claims 1-7.

10. A computer storage medium comprising instructions which, when run on a computer, cause the computer to perform the method of any one of claims 1-7.