CN118964586B - Interactive data processing method and system based on text classification - Google Patents

Abstract

The invention discloses an interactive data processing method and system based on text classification, wherein the method comprises the steps of obtaining dialogue text sent by a target user to a target virtual role; the method comprises the steps of training a text classifier, determining a plurality of core texts and corresponding text parameters in a dialogue text according to the training, screening a knowledge graph model corresponding to a target virtual role in a candidate role knowledge graph base according to the plurality of core texts and the corresponding text parameters, inputting the plurality of core texts and the corresponding text parameters into the knowledge graph model to predict and obtain answer texts corresponding to the target virtual role, and displaying the answer texts to a target user. Therefore, the invention can provide more intelligent and various interactive dialogue services for users, improve the efficiency and the degree of freedom of the interactive dialogue services and reduce errors.

Description

Interactive data processing method and system based on text classification

Technical Field

The invention relates to the technical field of data processing, in particular to an interactive data processing method and system based on text classification.

Background

Currently, when man-machine interaction in a game service or a software product is implemented, most of the man-machine interaction is still implemented by adopting preset dialogue options and response rules, for example, when a player interacts with a virtual character, a plurality of options are provided for the player as speaking options, and answers corresponding to each option and subsequent further data rules are prestored on a server. This prior art approach obviously fails to provide a more intelligent dialogue interaction service for the user, and since the preset rules may have contradictions caused by insufficient tests, and also easily cause the interaction dialogue of the user to be in error, resulting in service breakdown, the stability of the dialogue interaction service is also problematic. It can be seen that the prior art has defects and needs to be solved.

Disclosure of Invention

The invention aims to solve the technical problem of providing an interactive data processing method and system based on text classification, which can provide more intelligent and diversified interactive dialogue services for users, improve the efficiency and the freedom degree of the interactive dialogue services and reduce errors.

To solve the above technical problem, the first aspect of the present invention discloses an interactive data processing method based on text classification, the method comprising:

acquiring a dialogue text sent by a target user to a target virtual character;

according to the trained text classifier, determining a plurality of core texts and corresponding text parameters in the dialogue text;

Screening a knowledge graph model corresponding to the target virtual character from a candidate character knowledge graph base according to the plurality of core texts and the corresponding text parameters;

And inputting the plurality of core texts and the corresponding text parameters into the knowledge graph model to predict and obtain answer texts corresponding to the target virtual roles for display to the target users.

As an optional implementation manner, in the first aspect of the present invention, the core text is a dialogue purpose text, a dialogue motivation text, a next action description text, a previous action summary text or an instruction text.

In a first aspect of the present invention, the text parameters include a text purpose and text security, the text purpose being to issue a question, answer a question, make a selection, end a selection, initiate an association, start a scenario, advance a scenario, or end a scenario.

As an optional implementation manner, in the first aspect of the present invention, the determining, according to the trained text classifier, a plurality of core texts and corresponding text parameters in the dialog text includes:

Word segmentation and random division are carried out on the dialogue text to obtain a plurality of text fragments;

Performing preliminary classification on all the text fragments based on a clustering algorithm to obtain a plurality of text fragment sets;

Inputting each text segment in each text segment set into a trained core text prediction neural network for obtaining core text probability corresponding to each text segment; the core text classification neural network is obtained through training a training data set comprising a plurality of training dialogue texts and corresponding core text labels;

calculating the average value of the core text probabilities of all the text fragments in the text fragment set to obtain a set probability parameter corresponding to the text fragment set;

for each text segment, calculating the product of the core text probability corresponding to the text segment and the corresponding probability weight to obtain the segment probability parameter corresponding to the text segment, wherein the probability weight is in direct proportion to the set probability parameter corresponding to the text segment set to which the text segment belongs;

determining the text segment with the segment probability parameter larger than a first parameter threshold as a core text;

And determining the text parameters corresponding to each core text according to the text fragment set corresponding to the core text and the text parameter identification neural network.

In a first aspect of the present invention, as an optional implementation manner, the performing preliminary classification on all the text segments based on a clustering algorithm to obtain a plurality of text segment sets includes:

setting an objective function to enable the number of texts in each text fragment set to be the largest and the total set number of all the text fragment sets to be the smallest;

The setting of the limiting conditions includes:

The text similarity between any text segment in each text segment set and the text formed by all text segments in the text segment set is greater than a first similarity threshold;

The text similarity between the texts of all text segment combinations of any two different text segment sets is smaller than a second similarity threshold value, wherein the second similarity threshold value is smaller than the first similarity threshold value;

The membership parameter between any text segment in each text segment set and all text segments in the text segment set is greater than a first membership threshold;

The membership parameter between any text segment in each text segment set and all text segments in any other text segment set is smaller than a second membership threshold, wherein the second membership threshold is smaller than the first membership threshold;

and based on a dynamic programming algorithm, carrying out iterative classification on all the text fragments based on a clustering algorithm according to the objective function and the limiting condition until convergence, and obtaining a plurality of text fragment sets.

As an optional implementation manner, in the first aspect of the present invention, the determining, according to the text segment set corresponding to the core text and the text parameter identification neural network, a text parameter corresponding to each core text includes:

for each core text, determining a text segment with a probability parameter greater than a second parameter threshold value in the text segment set corresponding to the core text as a relevant text segment;

Inputting each relevant text segment corresponding to the core text into a trained text parameter identification neural network to obtain a predicted text parameter corresponding to each relevant text segment, wherein the text parameter identification neural network is trained by a training data set comprising a plurality of training texts and corresponding text parameter labels;

Calculating intersections among the predicted text destination parameters corresponding to all the related text fragments to obtain text destinations corresponding to the core text;

And calculating the average value of the predicted text safety corresponding to all the related text fragments to obtain the text safety corresponding to the core text.

In a first aspect of the present invention, the selecting, according to the plurality of core texts and the corresponding text parameters, the knowledge-graph model corresponding to the target virtual character from the candidate character knowledge-graph library includes:

For each candidate knowledge graph model in the candidate character knowledge graph library, acquiring training knowledge base data corresponding to the candidate knowledge graph model;

analyzing the training knowledge base data to obtain a corresponding knowledge base role parameter set and knowledge base text parameter set;

Calculating the model matching degree between the candidate knowledge graph model and the target virtual character according to the plurality of core texts and the corresponding text parameters, the knowledge base character parameter set and the knowledge base text parameter set;

and determining the candidate knowledge graph model with the highest model matching degree as the knowledge graph model corresponding to the target virtual role.

As an optional implementation manner, in the first aspect of the present invention, the calculating a model matching degree between the candidate knowledge-graph model and the target virtual character according to the plurality of core texts and the corresponding text parameters, the knowledge-base role parameter set and the knowledge-base text parameter set includes:

calculating an average value of first similarity between the text parameter corresponding to each core text and the knowledge base text parameter set to obtain a first matching degree parameter;

Calculating a second similarity between the character parameter of the target virtual character and the character parameter set of the knowledge base, wherein the character parameter or the character parameter set of the knowledge base comprises a character name, a character profile, a character occupation, a character function and a character gender;

And calculating the product of the first matching degree parameter and the second similarity to obtain the model matching degree between the candidate knowledge graph model and the target virtual role.

A second aspect of an embodiment of the present invention discloses an interactive data processing system based on text classification, the system comprising:

the acquisition module is used for acquiring a dialogue text sent to the target virtual role by the target user;

The classification module is used for determining a plurality of core texts and corresponding text parameters in the dialogue texts according to the trained text classifier;

The screening module is used for screening the knowledge graph model corresponding to the target virtual character from the candidate character knowledge graph library according to the plurality of core texts and the corresponding text parameters;

and the prediction module is used for inputting the plurality of core texts and the corresponding text parameters into the knowledge graph model so as to obtain the answer text corresponding to the target virtual character in a prediction mode, and displaying the answer text to the target user.

As an alternative embodiment, in the second aspect of the present invention, the core text is a dialogue destination text, a dialogue motivation text, a next action description text, a previous action summary text, or an instruction text.

In a second aspect of the invention, the text parameters include a text purpose and text security, the text purpose being to issue a question, answer a question, make a selection, end a selection, initiate an association, start a scenario, advance a scenario, or end a scenario.

In a second aspect of the present invention, the classifying module determines, according to a trained text classifier, a specific manner of determining a plurality of core texts and corresponding text parameters in the dialog text, including:

Word segmentation and random division are carried out on the dialogue text to obtain a plurality of text fragments;

Performing preliminary classification on all the text fragments based on a clustering algorithm to obtain a plurality of text fragment sets;

Inputting each text segment in each text segment set into a trained core text prediction neural network for obtaining core text probability corresponding to each text segment; the core text classification neural network is obtained through training a training data set comprising a plurality of training dialogue texts and corresponding core text labels;

calculating the average value of the core text probabilities of all the text fragments in the text fragment set to obtain a set probability parameter corresponding to the text fragment set;

for each text segment, calculating the product of the core text probability corresponding to the text segment and the corresponding probability weight to obtain the segment probability parameter corresponding to the text segment, wherein the probability weight is in direct proportion to the set probability parameter corresponding to the text segment set to which the text segment belongs;

determining the text segment with the segment probability parameter larger than a first parameter threshold as a core text;

And determining the text parameters corresponding to each core text according to the text fragment set corresponding to the core text and the text parameter identification neural network.

In a second aspect of the present invention, as an optional implementation manner, the specific manner in which the classification module performs preliminary classification on all the text segments based on a clustering algorithm to obtain a plurality of text segment sets includes:

setting an objective function to enable the number of texts in each text fragment set to be the largest and the total set number of all the text fragment sets to be the smallest;

The setting of the limiting conditions includes:

The text similarity between any text segment in each text segment set and the text formed by all text segments in the text segment set is greater than a first similarity threshold;

The text similarity between the texts of all text segment combinations of any two different text segment sets is smaller than a second similarity threshold value, wherein the second similarity threshold value is smaller than the first similarity threshold value;

The membership parameter between any text segment in each text segment set and all text segments in the text segment set is greater than a first membership threshold;

The membership parameter between any text segment in each text segment set and all text segments in any other text segment set is smaller than a second membership threshold, wherein the second membership threshold is smaller than the first membership threshold;

and based on a dynamic programming algorithm, carrying out iterative classification on all the text fragments based on a clustering algorithm according to the objective function and the limiting condition until convergence, and obtaining a plurality of text fragment sets.

As an optional implementation manner, in the second aspect of the present invention, the determining, by the classification module, a specific manner of determining a text parameter corresponding to each core text according to the text segment set corresponding to the core text and a text parameter identification neural network includes:

for each core text, determining a text segment with a probability parameter greater than a second parameter threshold value in the text segment set corresponding to the core text as a relevant text segment;

Inputting each relevant text segment corresponding to the core text into a trained text parameter identification neural network to obtain a predicted text parameter corresponding to each relevant text segment, wherein the text parameter identification neural network is trained by a training data set comprising a plurality of training texts and corresponding text parameter labels;

Calculating intersections among the predicted text destination parameters corresponding to all the related text fragments to obtain text destinations corresponding to the core text;

And calculating the average value of the predicted text safety corresponding to all the related text fragments to obtain the text safety corresponding to the core text.

In a second aspect of the present invention, the specific manner of screening the knowledge-graph model corresponding to the target virtual character from the candidate character knowledge-graph library according to the plurality of core texts and the corresponding text parameters includes:

For each candidate knowledge graph model in the candidate character knowledge graph library, acquiring training knowledge base data corresponding to the candidate knowledge graph model;

analyzing the training knowledge base data to obtain a corresponding knowledge base role parameter set and knowledge base text parameter set;

Calculating the model matching degree between the candidate knowledge graph model and the target virtual character according to the plurality of core texts and the corresponding text parameters, the knowledge base character parameter set and the knowledge base text parameter set;

and determining the candidate knowledge graph model with the highest model matching degree as the knowledge graph model corresponding to the target virtual role.

As an optional implementation manner, in the second aspect of the present invention, the specific manner of calculating, by the screening module, the model matching degree between the candidate knowledge-graph model and the target virtual character according to the plurality of core texts and the corresponding text parameters, the knowledge-base role parameter set and the knowledge-base text parameter set includes:

calculating an average value of first similarity between the text parameter corresponding to each core text and the knowledge base text parameter set to obtain a first matching degree parameter;

Calculating a second similarity between the character parameter of the target virtual character and the character parameter set of the knowledge base, wherein the character parameter or the character parameter set of the knowledge base comprises a character name, a character profile, a character occupation, a character function and a character gender;

And calculating the product of the first matching degree parameter and the second similarity to obtain the model matching degree between the candidate knowledge graph model and the target virtual role.

A third aspect of the invention discloses another interactive data processing system based on text classification, said system comprising:

a memory storing executable program code;

a processor coupled to the memory;

The processor invokes the executable program code stored in the memory to perform some or all of the steps in the text classification based interactive data processing method disclosed in the first aspect of the invention.

A fourth aspect of the invention discloses a computer storage medium storing computer instructions which, when invoked, are adapted to perform part or all of the steps of the text classification based interactive data processing method disclosed in the first aspect of the invention.

Compared with the prior art, the embodiment of the invention has the following beneficial effects:

the method and the device can determine a plurality of core texts and corresponding text parameters in the dialogue texts of the user based on the trained text classifier, and screen out a proper knowledge-graph model based on the core texts and the corresponding text parameters to predict reasonable and accurate answer texts, so that more intelligent and various interactive dialogue services can be provided for the user, the efficiency and the freedom degree of the interactive dialogue services are improved, and errors are reduced.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 is a schematic flow chart of an interactive data processing method based on text classification according to an embodiment of the present invention.

FIG. 2 is a schematic diagram of an interactive data processing system based on text classification in accordance with an embodiment of the present invention.

FIG. 3 is a schematic diagram of another interactive data processing system based on text classification in accordance with an embodiment of the present invention.

Detailed Description

In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

The terms first, second and the like in the description and in the claims and in the above-described figures are used for distinguishing between different objects and not necessarily for describing a sequential or chronological order. Furthermore, the terms "comprise" and "have," as well as any variations thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, apparatus, article, or device that comprises a list of steps or elements is not limited to the list of steps or elements but may, in the alternative, include other steps or elements not expressly listed or inherent to such process, method, article, or device.

Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the invention. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Those of skill in the art will explicitly and implicitly appreciate that the embodiments described herein may be combined with other embodiments.

The invention discloses an interactive data processing method and system based on text classification, which can determine a plurality of core texts and corresponding text parameters in dialogue texts of users based on a trained text classifier, and screen out a proper knowledge graph model based on the core texts and the corresponding text parameters to predict reasonable and accurate answer texts, so that more intelligent and various interactive dialogue services can be provided for the users, the efficiency and the degree of freedom of the interactive dialogue services are improved, and errors are reduced. The following will describe in detail.

Example 1

Referring to fig. 1, fig. 1 is a flow chart of an interactive data processing method based on text classification according to an embodiment of the present invention. The interactive data processing method based on text classification described in fig. 1 can be applied to a data processing system/data processing device/data processing server (wherein the server comprises a local processing server or a cloud processing server). As shown in fig. 1, the text classification-based interactive data processing method may include the operations of:

101. And acquiring dialogue text sent by the target user to the target virtual character.

102. And determining a plurality of core texts and corresponding text parameters in the dialogue text according to the trained text classifier.

103. And screening the knowledge graph model corresponding to the target virtual character from the candidate character knowledge graph library according to the plurality of core texts and the corresponding text parameters.

104. And inputting the plurality of core texts and the corresponding text parameters into the knowledge graph model to predict and obtain answer texts corresponding to the target virtual roles for display to target users.

Therefore, the embodiment of the invention can determine a plurality of core texts and corresponding text parameters in the dialogue texts of the user based on the trained text classifier, and screen out a proper knowledge-graph model based on the core texts and the corresponding text parameters to predict a reasonable and accurate answer text, so that more intelligent and various interactive dialogue services can be provided for the user, the efficiency and the freedom degree of the interactive dialogue services are improved, and errors are reduced.

As an alternative embodiment, in the above steps, the core text is dialogue purpose text, dialogue motivation text, next action description text, last action summary text, or instruction text.

Therefore, through the optional embodiment, the types of the core texts are clarified, the characteristics of the core texts in the dialogue texts of the user can be more accurately represented, so that a proper knowledge graph model is accurately screened out, reasonable and accurate answer texts are predicted, the intelligent and various interactive dialogue services are provided for the user in an auxiliary mode, the efficiency and the freedom degree of the interactive dialogue services are improved, and errors are reduced.

As an alternative embodiment, the text parameters in the steps above include text intent and text security, the text intent being to issue a question, answer a question, make a selection, end a selection, initiate an association, open a scenario, advance a scenario, or end a scenario.

In a specific implementation idea, in the step of predicting in step 104, the core text with the text security lower than the preset security threshold is removed, or is input into the knowledge-graph model after being specially marked, so as to specifically require the knowledge-graph model to perform sensitivity inspection or shielding on the core text, or further determine whether the target user is malicious or not based on the core text.

Therefore, through the optional embodiment, the types of the text parameters are clarified, the text object characteristics and the security characteristics in the dialogue text of the user can be more accurately represented, so that a proper knowledge graph model is accurately screened out, a reasonable and accurate answer text is predicted, more intelligent and various interactive dialogue services are provided for the user in an auxiliary mode, the efficiency and the freedom degree of the interactive dialogue services are improved, and errors are reduced.

As an optional embodiment, in the step, determining, according to the trained text classifier, a plurality of core texts and corresponding text parameters in the dialog text includes:

word segmentation and random division are carried out on the dialogue text to obtain a plurality of text fragments;

Performing preliminary classification on all the text fragments based on a clustering algorithm to obtain a plurality of text fragment sets;

For each text segment set, inputting each text segment in the text segment set into a trained core text prediction neural network to obtain a core text probability corresponding to each text segment; optionally, the core text classification neural network is obtained by training a training data set comprising a plurality of training dialogue texts and corresponding core text labels;

calculating the average value of the core text probabilities of all the text fragments in the text fragment set to obtain a set probability parameter corresponding to the text fragment set;

For each text segment, calculating the product of the core text probability corresponding to the text segment and the corresponding probability weight to obtain the segment probability parameter corresponding to the text segment;

determining a text segment with the segment probability parameter larger than a first parameter threshold as a core text;

And identifying the neural network according to the text fragment set corresponding to the core text and the text parameters, and determining the text parameters corresponding to each core text.

Therefore, through the above-mentioned alternative embodiment, classification, recognition and prediction can be carried out layer by layer on the dialogue text through the trained core text classification neural network and text parameter recognition neural network, so as to obtain accurate core text and corresponding text parameters, and the appropriate knowledge graph model can be conveniently and accurately screened out and the reasonable and accurate answer text can be predicted conveniently and accurately, so that the interactive dialogue service providing more intelligent and various interactive dialogue services for users can be assisted, the efficiency and the degree of freedom of the interactive dialogue service can be improved, and errors can be reduced.

As an optional embodiment, in the step, performing preliminary classification on all text segments based on a clustering algorithm to obtain a plurality of text segment sets, including:

Setting an objective function to enable the number of texts in each text fragment set to be the largest and the total set number of all text fragment sets to be the smallest;

The setting of the limiting conditions includes:

The text similarity between any text segment in each text segment set and the text formed by all text segments in the text segment set is greater than a first similarity threshold;

The text similarity between the texts of all text segment combinations of any two different text segment sets is less than a second similarity threshold, optionally the second similarity threshold is less than the first similarity threshold;

The membership parameter between any text segment in each text segment set and all text segments in the text segment set is greater than a first membership threshold;

the membership parameter between any text segment in each text segment set and all text segments in any other text segment set is smaller than a second membership threshold;

And based on a dynamic programming algorithm, carrying out iterative classification on all the text fragments based on a clustering algorithm according to the objective function and the limiting condition until convergence, and obtaining a plurality of text fragment sets.

Specifically, the technical thought of the classification algorithm combining the similarity and the membership degree can be referred to the scheme in the patent application of the invention with the patent application number of CN201610380813.7, where the specific calculation mode of the similarity and the membership degree can also be referred to the calculation thought in the patent of the invention, or be implemented by other existing related algorithms.

Therefore, through the above-mentioned alternative embodiment, through the classification objective function and the constraint condition that have been set up and combined similarity and membership, based on the dynamic programming algorithm, the text segment is initially classified, so as to obtain accurate core text and corresponding text parameters later, so that a proper knowledge graph model can be conveniently and accurately screened out, and reasonable and accurate answer text can be predicted, so that the interactive dialogue service can be provided for users more intelligently and variously, the efficiency and the freedom of the interactive dialogue service can be improved, and errors can be reduced.

As an optional embodiment, in the step, the determining the text parameter corresponding to each core text according to the text segment set corresponding to the core text and the text parameter identification neural network includes:

For each core text, determining the text segment with the probability parameter of each segment greater than a second parameter threshold value in the text segment set corresponding to the core text as a related text segment;

Inputting each relevant text segment corresponding to the core text into a trained text parameter identification neural network to obtain a predicted text parameter corresponding to each relevant text segment, wherein the text parameter identification neural network is obtained by training a training data set comprising a plurality of training texts and corresponding text parameter labels;

calculating intersections among the predicted text destination parameters corresponding to all the related text fragments to obtain text destinations corresponding to the core text;

and calculating the average value of the predicted text safety corresponding to all the related text fragments to obtain the text safety corresponding to the core text.

Therefore, through the above-mentioned alternative embodiment, through the screening of the relevant text fragments in the text fragment set corresponding to the core text, and through the trained text parameter recognition neural network, the text parameter recognition and intersection average calculation can be realized, so as to obtain the accurate core text and the corresponding text parameters, so that the appropriate knowledge graph model can be conveniently and accurately screened out, the reasonable and accurate answer text can be predicted, the more intelligent and various interactive dialogue services can be provided for the user in an auxiliary manner, the efficiency and the degree of freedom of the interactive dialogue service can be improved, and the errors can be reduced.

As an optional embodiment, in the step, according to the multiple core texts and the corresponding text parameters, a knowledge graph model corresponding to the target virtual character is screened out from the candidate character knowledge graph library, including:

For each candidate knowledge graph model in the candidate character knowledge graph library, acquiring training knowledge base data corresponding to the candidate knowledge graph model;

Analyzing training knowledge base data to obtain a corresponding knowledge base role parameter set and knowledge base text parameter set;

calculating the model matching degree between the candidate knowledge graph model and the target virtual character according to the plurality of core texts and the corresponding text parameters, the knowledge base character parameter set and the knowledge base text parameter set;

and determining the candidate knowledge graph model with the highest model matching degree as the knowledge graph model corresponding to the target virtual role.

Specifically, all character parameter data and training text data in the training knowledge base data can be counted to obtain a corresponding knowledge base character parameter set and a knowledge base text parameter set.

Therefore, through the above optional embodiment, the corresponding data parameter set can be determined through analysis of the knowledge base data of the candidate knowledge graph model, and the knowledge graph model corresponding to the target virtual role is determined based on calculation and screening of the matching degree, so as to assist in providing more intelligent and various interactive dialogue services for users, improve the efficiency and the freedom degree of the interactive dialogue services, and reduce errors.

As an optional embodiment, in the step, calculating the model matching degree between the candidate knowledge-graph model and the target virtual character according to the plurality of core texts and the corresponding text parameters, the knowledge-base character parameter set and the knowledge-base text parameter set includes:

Calculating an average value of first similarity between text parameters corresponding to each core text and a text parameter set of a knowledge base to obtain a first matching degree parameter;

Optionally, the character parameter or the knowledge base character parameter set comprises a character name, a character profile, a character occupation, a character function and a character gender;

and calculating the product of the first matching degree parameter and the second similarity to obtain the model matching degree between the candidate knowledge graph model and the target virtual role.

Therefore, according to the above-mentioned alternative embodiment, the matching degree of each candidate model can be obtained by calculating the similarity between the role parameter and the text parameter of the knowledge base data of the candidate knowledge base model, the text parameter of the current dialogue text and the role parameter of the current target virtual role, so that the knowledge base model corresponding to the target virtual role can be accurately determined based on the matching degree, thereby assisting in providing more intelligent and diversified interactive dialogue services for users, improving the efficiency and the freedom degree of the interactive dialogue service, and reducing errors.

Example two

Referring now to FIG. 2, FIG. 2 is a schematic diagram illustrating an interactive data processing system based on text classification according to an embodiment of the present invention. The text classification based interactive data processing system depicted in fig. 2 may be applied, among other things, in a data processing system/data processing device/data processing server (where the server includes a local processing server or a cloud processing server). As shown in fig. 2, the text classification based interactive data processing system may include:

and the acquisition module 201 is used for acquiring the dialogue text sent to the target virtual character by the target user.

The classification module 202 is configured to determine a plurality of core texts and corresponding text parameters in the dialog text according to the trained text classifier.

And the screening module 203 is configured to screen a knowledge graph model corresponding to the target virtual character from the candidate character knowledge graph library according to the plurality of core texts and the corresponding text parameters.

And the prediction module 204 is configured to input a plurality of core texts and corresponding text parameters into the knowledge graph model to predict and obtain an answer text corresponding to the target virtual character for display to the target user.

Therefore, the embodiment of the invention can determine a plurality of core texts and corresponding text parameters in the dialogue texts of the user based on the trained text classifier, and screen out a proper knowledge-graph model based on the core texts and the corresponding text parameters to predict a reasonable and accurate answer text, so that more intelligent and various interactive dialogue services can be provided for the user, the efficiency and the freedom degree of the interactive dialogue services are improved, and errors are reduced.

As an alternative embodiment, the core text is dialogue purpose text, dialogue motivation text, next action description text, last action summary text, or instruction text.

Therefore, through the optional embodiment, the types of the core texts are clarified, the characteristics of the core texts in the dialogue texts of the user can be more accurately represented, so that a proper knowledge graph model is accurately screened out, reasonable and accurate answer texts are predicted, the intelligent and various interactive dialogue services are provided for the user in an auxiliary mode, the efficiency and the freedom degree of the interactive dialogue services are improved, and errors are reduced.

As an alternative embodiment, the text parameters include text intent and text security, the text intent being to issue a question, answer a question, make a selection, end a selection, initiate an association, open a scenario, advance a scenario, or end a scenario.

Therefore, through the optional embodiment, the types of the text parameters are clarified, the text object characteristics and the security characteristics in the dialogue text of the user can be more accurately represented, so that a proper knowledge graph model is accurately screened out, a reasonable and accurate answer text is predicted, more intelligent and various interactive dialogue services are provided for the user in an auxiliary mode, the efficiency and the freedom degree of the interactive dialogue services are improved, and errors are reduced.

As an optional embodiment, the classifying module determines, according to the trained text classifier, a specific manner of a plurality of core texts and corresponding text parameters in the dialog text, including:

word segmentation and random division are carried out on the dialogue text to obtain a plurality of text fragments;

Performing preliminary classification on all the text fragments based on a clustering algorithm to obtain a plurality of text fragment sets;

For each text segment set, inputting each text segment in the text segment set into a trained core text prediction neural network to obtain a core text probability corresponding to each text segment; optionally, the core text classification neural network is obtained by training a training data set comprising a plurality of training dialogue texts and corresponding core text labels;

calculating the average value of the core text probabilities of all the text fragments in the text fragment set to obtain a set probability parameter corresponding to the text fragment set;

For each text segment, calculating the product of the core text probability corresponding to the text segment and the corresponding probability weight to obtain the segment probability parameter corresponding to the text segment;

determining a text segment with the segment probability parameter larger than a first parameter threshold as a core text;

And identifying the neural network according to the text fragment set corresponding to the core text and the text parameters, and determining the text parameters corresponding to each core text.

Therefore, through the above-mentioned alternative embodiment, classification, recognition and prediction can be carried out layer by layer on the dialogue text through the trained core text classification neural network and text parameter recognition neural network, so as to obtain accurate core text and corresponding text parameters, and the appropriate knowledge graph model can be conveniently and accurately screened out and the reasonable and accurate answer text can be predicted conveniently and accurately, so that the interactive dialogue service providing more intelligent and various interactive dialogue services for users can be assisted, the efficiency and the degree of freedom of the interactive dialogue service can be improved, and errors can be reduced.

As an optional embodiment, the specific manner of performing preliminary classification on all text segments based on a clustering algorithm by the classification module to obtain a plurality of text segment sets includes:

Setting an objective function to enable the number of texts in each text fragment set to be the largest and the total set number of all text fragment sets to be the smallest;

The setting of the limiting conditions includes:

The text similarity between any text segment in each text segment set and the text formed by all text segments in the text segment set is greater than a first similarity threshold;

The text similarity between the texts of all text segment combinations of any two different text segment sets is less than a second similarity threshold, optionally the second similarity threshold is less than the first similarity threshold;

The membership parameter between any text segment in each text segment set and all text segments in the text segment set is greater than a first membership threshold;

the membership parameter between any text segment in each text segment set and all text segments in any other text segment set is smaller than a second membership threshold;

And based on a dynamic programming algorithm, carrying out iterative classification on all the text fragments based on a clustering algorithm according to the objective function and the limiting condition until convergence, and obtaining a plurality of text fragment sets.

Therefore, through the above-mentioned alternative embodiment, through the classification objective function and the constraint condition that have been set up and combined similarity and membership, based on the dynamic programming algorithm, the text segment is initially classified, so as to obtain accurate core text and corresponding text parameters later, so that a proper knowledge graph model can be conveniently and accurately screened out, and reasonable and accurate answer text can be predicted, so that the interactive dialogue service can be provided for users more intelligently and variously, the efficiency and the freedom of the interactive dialogue service can be improved, and errors can be reduced.

As an optional embodiment, the classifying module identifies the neural network according to the text segment set corresponding to the core text and the text parameters, and determines a specific mode of the text parameters corresponding to each core text, including:

For each core text, determining the text segment with the probability parameter of each segment greater than a second parameter threshold value in the text segment set corresponding to the core text as a related text segment;

Inputting each relevant text segment corresponding to the core text into a trained text parameter identification neural network to obtain a predicted text parameter corresponding to each relevant text segment, wherein the text parameter identification neural network is obtained by training a training data set comprising a plurality of training texts and corresponding text parameter labels;

calculating intersections among the predicted text destination parameters corresponding to all the related text fragments to obtain text destinations corresponding to the core text;

and calculating the average value of the predicted text safety corresponding to all the related text fragments to obtain the text safety corresponding to the core text.

Therefore, through the above-mentioned alternative embodiment, through the screening of the relevant text fragments in the text fragment set corresponding to the core text, and through the trained text parameter recognition neural network, the text parameter recognition and intersection average calculation can be realized, so as to obtain the accurate core text and the corresponding text parameters, so that the appropriate knowledge graph model can be conveniently and accurately screened out, the reasonable and accurate answer text can be predicted, the more intelligent and various interactive dialogue services can be provided for the user in an auxiliary manner, the efficiency and the degree of freedom of the interactive dialogue service can be improved, and the errors can be reduced.

As an optional embodiment, the specific way of screening the knowledge spectrum model corresponding to the target virtual character from the candidate character knowledge spectrum library by the screening module according to the plurality of core texts and the corresponding text parameters includes:

For each candidate knowledge graph model in the candidate character knowledge graph library, acquiring training knowledge base data corresponding to the candidate knowledge graph model;

Analyzing training knowledge base data to obtain a corresponding knowledge base role parameter set and knowledge base text parameter set;

calculating the model matching degree between the candidate knowledge graph model and the target virtual character according to the plurality of core texts and the corresponding text parameters, the knowledge base character parameter set and the knowledge base text parameter set;

and determining the candidate knowledge graph model with the highest model matching degree as the knowledge graph model corresponding to the target virtual role.

Therefore, through the above optional embodiment, the corresponding data parameter set can be determined through analysis of the knowledge base data of the candidate knowledge graph model, and the knowledge graph model corresponding to the target virtual role is determined based on calculation and screening of the matching degree, so as to assist in providing more intelligent and various interactive dialogue services for users, improve the efficiency and the freedom degree of the interactive dialogue services, and reduce errors.

As an optional embodiment, the specific manner of calculating the model matching degree between the candidate knowledge-graph model and the target virtual character according to the plurality of core texts and the corresponding text parameters, the knowledge-base role parameter set and the knowledge-base text parameter set by the screening module includes:

Calculating an average value of first similarity between text parameters corresponding to each core text and a text parameter set of a knowledge base to obtain a first matching degree parameter;

Optionally, the character parameter or the knowledge base character parameter set comprises a character name, a character profile, a character occupation, a character function and a character gender;

and calculating the product of the first matching degree parameter and the second similarity to obtain the model matching degree between the candidate knowledge graph model and the target virtual role.

Therefore, according to the above-mentioned alternative embodiment, the matching degree of each candidate model can be obtained by calculating the similarity between the role parameter and the text parameter of the knowledge base data of the candidate knowledge base model, the text parameter of the current dialogue text and the role parameter of the current target virtual role, so that the knowledge base model corresponding to the target virtual role can be accurately determined based on the matching degree, thereby assisting in providing more intelligent and diversified interactive dialogue services for users, improving the efficiency and the freedom degree of the interactive dialogue service, and reducing errors.

Example III

Referring now to FIG. 3, FIG. 3 is a diagram illustrating yet another interactive data processing system based on text classification in accordance with an embodiment of the present invention. The text classification based interactive data processing system depicted in fig. 3 is applied in a data processing system/data processing device/data processing server, wherein the server comprises a local processing server or a cloud processing server. As shown in fig. 3, the text classification based interactive data processing system may include:

a memory 301 storing executable program code;

A processor 302 coupled with the memory 301;

wherein the processor 302 invokes executable program code stored in the memory 301 for performing the steps of the text classification based interactive data processing method described in embodiment one.

Example IV

The embodiment of the invention discloses a computer-readable storage medium storing a computer program for electronic data exchange, wherein the computer program causes a computer to execute the steps of the interactive data processing method based on text classification described in the embodiment.

Example five

The present invention discloses a computer program product comprising a non-transitory computer readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps of the interactive data processing method based on text classification described in the embodiment.

The foregoing describes certain embodiments of the present disclosure, other embodiments being within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve desirable results. Furthermore, the processes depicted in the accompanying drawings do not necessarily have to be in the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

The system, apparatus, module or unit set forth in the above embodiments may be implemented in particular by a computer chip or entity, or by a product having a certain function. One typical implementation is a computer. In particular, the computer may be, for example, a personal computer, a laptop computer, a cellular telephone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

For convenience of description, the above devices are described as being functionally divided into various units, respectively. Of course, the functions of each element may be implemented in one or more software and/or hardware elements when implemented in the present specification.

It will be appreciated by those skilled in the art that the present description may be provided as a method, system, or computer program product. Accordingly, the present specification embodiments may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present description embodiments may take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

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

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

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

In one typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include volatile memory in a computer-readable medium, random Access Memory (RAM) and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of computer-readable media.

Computer readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Disks (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. Computer-readable media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises an element.

The description may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.

In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. In particular, for system embodiments, since they are substantially similar to method embodiments, the description is relatively simple, as relevant to see a section of the description of method embodiments.

Finally, it should be noted that the disclosure of the text classification-based interactive data processing method and system in the embodiments of the present invention is only a preferred embodiment of the present invention, and is only for illustrating the technical solution of the present invention, but not for limiting the same, and although the present invention has been described in detail with reference to the foregoing embodiments, it should be understood by those skilled in the art that the technical solution described in the foregoing embodiments may be modified or some of the technical features thereof may be equivalently replaced, and these modifications or replacements do not make the essence of the corresponding technical solution deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (8)

1. A method of interactive data processing based on text classification, the method comprising:

acquiring a dialogue text sent by a target user to a target virtual character;

according to the trained text classifier, determining a plurality of core texts and corresponding text parameters in the dialogue text;

Screening a knowledge graph model corresponding to the target virtual character from a candidate character knowledge graph library according to the plurality of core texts and the corresponding text parameters, wherein the method comprises the following steps:

For each candidate knowledge graph model in the candidate character knowledge graph library, acquiring training knowledge base data corresponding to the candidate knowledge graph model;

analyzing the training knowledge base data to obtain a corresponding knowledge base role parameter set and knowledge base text parameter set;

calculating an average value of first similarity between the text parameter corresponding to each core text and the knowledge base text parameter set to obtain a first matching degree parameter;

Calculating a second similarity between the character parameter of the target virtual character and the character parameter set of the knowledge base, wherein the character parameter or the character parameter set of the knowledge base comprises a character name, a character profile, a character occupation, a character function and a character gender;

calculating the product of the first matching degree parameter and the second similarity to obtain the model matching degree between the candidate knowledge graph model and the target virtual role;

determining the candidate knowledge graph model with the highest model matching degree as a knowledge graph model corresponding to the target virtual role;

And inputting the plurality of core texts and the corresponding text parameters into the knowledge graph model to predict and obtain answer texts corresponding to the target virtual roles for display to the target users.

2. The interactive data processing method based on text classification as claimed in claim 1, wherein the core text is a dialogue purpose text, a dialogue engine text, a next action description text, a previous action summary text, or an instruction text.

3. The text-classification-based interactive data processing method of claim 1, wherein the text parameters include a text purpose and a text security, the text purpose being to question, answer a question, make a selection, end a selection, initiate an association, start a scenario, advance a scenario, or end a scenario.

4. A text-based interactive data processing method according to claim 3, wherein said determining a plurality of core texts and corresponding text parameters in said dialog text according to a trained text classifier comprises:

Word segmentation and random division are carried out on the dialogue text to obtain a plurality of text fragments;

Performing preliminary classification on all the text fragments based on a clustering algorithm to obtain a plurality of text fragment sets;

Inputting each text segment in each text segment set into a trained core text prediction neural network for obtaining core text probability corresponding to each text segment; the core text prediction neural network is obtained through training a training data set comprising a plurality of training dialogue texts and corresponding core text labels;

calculating the average value of the core text probabilities of all the text fragments in the text fragment set to obtain a set probability parameter corresponding to the text fragment set;

for each text segment, calculating the product of the core text probability corresponding to the text segment and the corresponding probability weight to obtain the segment probability parameter corresponding to the text segment, wherein the probability weight is in direct proportion to the set probability parameter corresponding to the text segment set to which the text segment belongs;

determining the text segment with the segment probability parameter larger than a first parameter threshold as a core text;

And determining the text parameters corresponding to each core text according to the text fragment set corresponding to the core text and the text parameter identification neural network.

5. The text classification based interactive data processing method according to claim 4, wherein said performing a preliminary classification on all the text segments based on a clustering algorithm to obtain a plurality of text segment sets comprises:

setting an objective function to enable the number of texts in each text fragment set to be the largest and the total set number of all the text fragment sets to be the smallest;

The setting of the limiting conditions includes:

The text similarity between any text segment in each text segment set and the text formed by all text segments in the text segment set is greater than a first similarity threshold;

The text similarity between the texts of all text segment combinations of any two different text segment sets is smaller than a second similarity threshold value, wherein the second similarity threshold value is smaller than the first similarity threshold value;

The membership parameter between any text segment in each text segment set and all text segments in the text segment set is greater than a first membership threshold;

The membership parameter between any text segment in each text segment set and all text segments in any other text segment set is smaller than a second membership threshold, wherein the second membership threshold is smaller than the first membership threshold;

and based on a dynamic programming algorithm, carrying out iterative classification on all the text fragments based on a clustering algorithm according to the objective function and the limiting condition until convergence, and obtaining a plurality of text fragment sets.

6. The text classification based interactive data processing method of claim 4, wherein said determining text parameters corresponding to each of said core texts based on said set of text segments corresponding to said core text and a text parameter identification neural network comprises:

for each core text, determining a text segment with a probability parameter greater than a second parameter threshold value in the text segment set corresponding to the core text as a relevant text segment;

Inputting each relevant text segment corresponding to the core text into a trained text parameter identification neural network to obtain a predicted text parameter corresponding to each relevant text segment, wherein the text parameter identification neural network is trained by a training data set comprising a plurality of training texts and corresponding text parameter labels;

Calculating intersections among the predicted text destination parameters corresponding to all the related text fragments to obtain text destinations corresponding to the core text;

And calculating the average value of the predicted text safety corresponding to all the related text fragments to obtain the text safety corresponding to the core text.

7. An interactive data processing system based on text classification, the system comprising:

the acquisition module is used for acquiring a dialogue text sent to the target virtual role by the target user;

The classification module is used for determining a plurality of core texts and corresponding text parameters in the dialogue texts according to the trained text classifier;

the screening module is configured to screen a knowledge graph model corresponding to the target virtual character from a candidate character knowledge graph library according to the multiple core texts and the corresponding text parameters, and includes:

For each candidate knowledge graph model in the candidate character knowledge graph library, acquiring training knowledge base data corresponding to the candidate knowledge graph model;

analyzing the training knowledge base data to obtain a corresponding knowledge base role parameter set and knowledge base text parameter set;

calculating an average value of first similarity between the text parameter corresponding to each core text and the knowledge base text parameter set to obtain a first matching degree parameter;

Calculating a second similarity between the character parameter of the target virtual character and the character parameter set of the knowledge base, wherein the character parameter or the character parameter set of the knowledge base comprises a character name, a character profile, a character occupation, a character function and a character gender;

calculating the product of the first matching degree parameter and the second similarity to obtain the model matching degree between the candidate knowledge graph model and the target virtual role;

determining the candidate knowledge graph model with the highest model matching degree as a knowledge graph model corresponding to the target virtual role;

and the prediction module is used for inputting the plurality of core texts and the corresponding text parameters into the knowledge graph model so as to obtain the answer text corresponding to the target virtual character in a prediction mode, and displaying the answer text to the target user.

8. An interactive data processing system based on text classification, the system comprising:

a memory storing executable program code;

a processor coupled to the memory;

The processor invokes the executable program code stored in the memory to perform the text classification based interactive data processing method of any of claims 1-6.