Intelligent interactive answering method and device for user questions and computer equipment

Abstract

The invention relates to an intelligent interactive answering method and device for user questions and computer equipment. The method comprises the steps of obtaining current problem information and current task content of a user, and identifying user appeal content corresponding to the current problem information; identifying correlation information between the user appeal content of the user and the current task content, and extracting task entity information of the current task content; when the task entity information of the current task content is not collected, generating a progressive entity question of the user, repeating the question and answer collection process, and stopping the iteration process until the task entity information of the current task content is collected; and generating question answering information of the user through a customer service strategy corresponding to the current task content based on the task entity information obtained by each iteration. By adopting the method, the intelligent interactive answering efficiency of the user questions can be improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method, device and computer equipment for intelligent interactive answering of user questions. Background Art

[0002] With the continuous development of conversational robots, a large number of enterprises have gradually introduced conversational robots to provide intelligent and efficient processing methods for customer service, call answering and other functions. In the traditional task-based multi-round dialogue system, the dialogue system is constructed by combining multiple independent modules (natural language processing (NLU), dialogue management (DM), dialogue generation (NLG)). Each module requires multiple models to support. Although this method can achieve complex dialogue management to a certain extent, it has many limitations that make it difficult to significantly improve the interaction effect. Therefore, how to improve the interaction effect of conversational robots is the current research focus.

[0003] Traditional conversational robot systems generally require the construction of multiple modules such as natural language processing (NLU), dialogue management (DM), and dialogue generation (NLG). Each module is trained by one or more models and then assembled together. However, this approach requires each module (NLU, DM, NLG) to be developed and optimized independently, which may lead to unnatural connections between modules, large systems, and complex models. The need to design and optimize each module separately makes the conversational robot system's understanding ability and interaction fluency poor, resulting in poor intelligent interactive answers to user questions. Summary of the invention

[0004] Based on this, it is necessary to provide a method, device, computer equipment, computer-readable storage medium and computer program product for intelligent interactive answering of user questions in response to the above technical problems.

[0005] In a first aspect, the present application provides a method for intelligent interactive answering of user questions, comprising:

[0006] Obtaining the user's current question information and the user's current task content, and identifying the user's demand content corresponding to the current question information;

[0007] Identify the correlation information between the user's user demand content and the user's current task content, and extract the task entity information of the current task content based on the user's user demand content when the correlation information meets a preset correlation condition;

[0008] When the task entity information of the current task content has not been collected, generating a progressive entity question for the user, and collecting the question answer content of the progressive entity question in response to the user's reply operation;

[0009] The question answer content replaces the user demand content, and returns to the step of extracting the task entity information of the current task content based on the user demand content, until the task entity information of the current task content is collected, and then stops the iteration process;

[0010] Based on the task entity information obtained in each iteration, the user's question answering information is generated through the customer service strategy corresponding to the current task content.

[0011] Optionally, the identifying the user demand content corresponding to the current question information includes:

[0012] Based on the current question information, identifying the question semantic content corresponding to the current question information through a semantic recognition network;

[0013] The text intention recognition network is used to identify the user question intention corresponding to the question semantic content, and the user question intention is used as the user demand content corresponding to the current question information.

[0014] Optionally, the identifying of the correlation information between the user's demand content and the user's current task content includes:

[0015] Based on the current task content of the user, extracting each prompt word information of the current task content through a keyword extraction strategy;

[0016] Based on the prompt word information and the user demand content, the similarity value between the user demand content and the prompt word information is identified through a semantic similarity recognition network, and the similarity value between the user demand content and the prompt word information is used as the correlation information between the user demand content of the user and the user's current task content.

[0017] Optionally, when the correlation information satisfies a preset correlation condition, extracting the task entity information of the current task content based on the user's user demand content includes:

[0018] Based on the similarity values ​​between the user demand content and each of the prompt word information, identifying a first number of prompt word information corresponding to a similarity value greater than a similarity threshold, and calculating a relevance value of the user demand content based on the first number and the total number of all prompt word information;

[0019] When the correlation value is greater than the correlation value threshold, it is determined that the correlation information meets the preset correlation condition, and the entity information extraction process is performed on the user demand content through the entity extraction model corresponding to the current task content to obtain the task entity information of the current task content.

[0020] Optionally, when the task entity information of the current task content has not been collected, generating the user's progressive entity question includes:

[0021] Based on the current task content, querying each task entity type corresponding to the current task content in a task database, and identifying each current task entity type contained in the task entity information of the current task content;

[0022] Based on each of the current task entity types, in each of the task entity types, each missing task entity type without entity content is screened, and in each of the missing task entity types, a target task entity type is randomly screened;

[0023] In the question database, the question content of the target task entity type is screened as the progressive entity question of the user.

[0024] Optionally, the generating of the user's question answering information based on the task entity information obtained in each iteration and through the customer service strategy corresponding to the current task content includes:

[0025] Based on the task entity information obtained in each iteration, identifying entity content information of each task entity type;

[0026] Based on the entity content information of each task entity type, identifying the task entity data of each task entity type, and based on the task entity data of each task entity type, querying each problem solution corresponding to the current problem information through the customer service strategy corresponding to the current task content;

[0027] All problem-solving solutions are regarded as the problem-solving information of the user.

[0028] In a second aspect, the present application also provides an intelligent interactive answering device for user questions, comprising:

[0029] An acquisition module is used to acquire the user's current question information and the user's current task content, and identify the user's demand content corresponding to the current question information;

[0030] An extraction module, used to identify the correlation information between the user's user demand content and the user's current task content, and extract the task entity information of the current task content based on the user's user demand content when the correlation information meets a preset correlation condition;

[0031] A collection module, configured to generate a progressive entity question for the user when the task entity information of the current task content has not been collected, and collect the question answer content of the progressive entity question in response to the user's reply operation;

[0032] An iteration module, used for replacing the user demand content with the question answer content, and returning to execute the step of extracting the task entity information of the current task content based on the user demand content, until the task entity information of the current task content is collected and the iteration process is stopped;

[0033] A generation module is used to generate the user's question answering information based on the task entity information obtained in each iteration and through the customer service strategy corresponding to the current task content.

[0034] Optionally, the acquisition module is specifically used to:

[0035] Based on the current question information, identifying the question semantic content corresponding to the current question information through a semantic recognition network;

[0036] The text intention recognition network is used to identify the user question intention corresponding to the question semantic content, and the user question intention is used as the user demand content corresponding to the current question information.

[0037] Optionally, the extraction module is specifically used to:

[0038] Based on the current task content of the user, extracting each prompt word information of the current task content through a keyword extraction strategy;

[0039] Based on the prompt word information and the user demand content, the similarity value between the user demand content and the prompt word information is identified through a semantic similarity recognition network, and the similarity value between the user demand content and the prompt word information is used as the correlation information between the user demand content of the user and the user's current task content.

[0040] Optionally, the extraction module is specifically used to:

[0041] Based on the similarity values ​​between the user demand content and each of the prompt word information, identifying a first number of prompt word information corresponding to a similarity value greater than a similarity threshold, and calculating a relevance value of the user demand content based on the first number and the total number of all prompt word information;

[0042] When the correlation value is greater than the correlation value threshold, it is determined that the correlation information meets the preset correlation condition, and the entity information extraction process is performed on the user demand content through the entity extraction model corresponding to the current task content to obtain the task entity information of the current task content.

[0043] Optionally, the acquisition module is specifically used to:

[0044] Based on the current task content, querying each task entity type corresponding to the current task content in a task database, and identifying each current task entity type contained in the task entity information of the current task content;

[0045] Based on each of the current task entity types, in each of the task entity types, each missing task entity type without entity content is screened, and in each of the missing task entity types, a target task entity type is randomly screened;

[0046] In the question database, the question content of the target task entity type is screened as the progressive entity question of the user.

[0047] Optionally, the generating module is specifically used for:

[0048] Based on the task entity information obtained in each iteration, identifying entity content information of each task entity type;

[0049] Based on the entity content information of each task entity type, identifying the task entity data of each task entity type, and based on the task entity data of each task entity type, querying each problem solution corresponding to the current problem information through the customer service strategy corresponding to the current task content;

[0050] All problem-solving solutions are regarded as the problem-solving information of the user.

[0051] In a third aspect, the present application provides a computer device, wherein the computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of any one of the methods in the first aspect are implemented.

[0052] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.

[0053] In a fifth aspect, the present application provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.

[0054] The above-mentioned intelligent interactive answering method, device and computer equipment for user questions obtain the user's current question information and the user's current task content, and identify the user demand content corresponding to the current question information; identify the correlation information between the user's user demand content and the user's current task content, and extract the task entity information of the current task content based on the user's user demand content when the correlation information meets the preset correlation condition; generate the user's progressive entity question when the task entity information of the current task content has not been collected, and collect the question answer content of the progressive entity question in response to the user's reply operation; replace the user demand content with the question answer content, and return to execute the step of extracting the task entity information of the current task content based on the user's user demand content, until the task entity information of the current task content is collected, and stop the iterative process; based on the task entity information obtained in each iteration, generate the user's question answering information through the customer service strategy corresponding to the current task content. This solution uses a large model to match the user's demand content of the user's question information with the current task content screened by the user, thereby extracting the task entity information related to the current task content in the user's question information, and when the entity information is not fully collected, through progressive question interaction, further collect the entity information to ensure that the entity information related to the current task content is fully collected. Finally, the terminal generates the user's question answer information based on the collected task entity information. It avoids the component stacking multi-module architecture of the traditional multi-round dialogue robot, and instead implements an adaptive multi-round dialogue robot and system structure based on a large model. It does not rely on traditional machine learning modules, but uses large models to complete all dialogue interactions, significantly improving the semantic understanding and dialogue logic coherence of the dialogue system, and improving the robot interaction effect. Thereby effectively improving the intelligent interactive answer effect to user questions. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0056] Figure 1A schematic diagram of a flow chart of a method for intelligent interactive answering of user questions in one embodiment;

[0057] Figure 2 A schematic diagram of the LoRA technology principle in one embodiment;

[0058] Figure 3 A flowchart of an example of intelligent interactive answering of user questions in one embodiment;

[0059] Figure 4 A structural block diagram of an intelligent interactive answering device for user questions in one embodiment;

[0060] Figure 5 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0062] The intelligent interactive answering method for user questions provided in the embodiment of the present application can be applied to the application environment of intelligent interactive answering of user questions. Among them, the method can be applied to a terminal, a server, or a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. Among them, the terminal can be, but is not limited to, various personal computers, laptops, medium-sized computers, etc. Among them, the terminal uses a large model to match the user's demand content of the user's question information with the current task content screened by the user, thereby extracting the task entity information related to the current task content in the user's question information, and when the entity information is not collected, through progressive question interaction, further collect each entity information to ensure that the entity information related to the current task content is collected. Finally, the terminal generates the user's question answering information based on the collected entity information of each task. It avoids the component stacking multi-module architecture of the traditional multi-round dialogue robot, but implements an adaptive multi-round dialogue robot and system structure based on a large model. It does not rely on traditional machine learning modules, uses large models to complete all dialogue interactions, significantly improves the semantic understanding of the dialogue system and the coherence of dialogue logic, and improves the robot interaction effect. Thereby effectively improving the intelligent interactive answering effect of user problems.

[0063] In an exemplary embodiment, Figure 1 As shown, a method for intelligent interactive answering of user questions is provided, which is described by taking the method applied to a terminal as an example, and includes the following steps S101 to S105. Among them:

[0064] Step S101, obtaining the user's current question information and the user's current task content, and identifying the user's demand content corresponding to the current question information.

[0065] In this embodiment, the terminal responds to the user's question check operation to obtain the user's current task content, wherein the current task content is the task information that the user needs to consult and communicate, such as life insurance reporting, bank transfer, insurance consultation and other task information, and then the terminal responds to the user's question input operation to obtain the user's current question information, wherein the input method of the current question information includes but is not limited to voice input, text input and other input methods. Then, the terminal identifies the user's demand content corresponding to the current question information. The specific identification process will be described in detail later.

[0066] Step S102, identifying the correlation information between the user's user demand content and the user's current task content, and extracting the task entity information of the current task content based on the user's user demand content when the correlation information meets a preset correlation condition.

[0067] In this embodiment, the terminal identifies the user's user demand content and the correlation information between the user's current task content, and extracts the task entity information of the current task content based on the user's user demand content when the correlation information meets the preset correlation condition. The preset correlation condition is whether the correlation value corresponding to the relevant information is greater than the correlation value threshold preset in the terminal. When it is greater than the correlation value threshold, it is determined that the correlation information meets the preset correlation condition. The specific determination process will be described in detail later. The method of extracting the task entity information is based on the following. Figure 2 The LoRA technology principle shown in the figure is based on the pre-trained model ChatGLM3-6B (6.2 billion parameters) for task entity information extraction. Among them, the model needs to be trained according to the sample information of different current task contents before application. The specific training process is:

[0068] 1) Prepare data: 50,000 hotline conversation data in the insurance field.

[0069] 2) Initialize the model: Initialize the parameters based on the pre-trained model ChatGLM3-6B (6.2 billion parameters). Zero initialization can be used.

[0070] 3) Model construction: Add a LoRA (Low-Rank Adaptation) layer on top of ChatGLM3-6B and fine-tune the model output by optimizing the parameters of this layer. The technical principle of LoRA is as follows Figure 1The core idea is to decompose the original model matrix into two low-rank matrices, reduce the dimension and then increase the dimension. When fine-tuning, the parameters of the pre-trained language model are fixed, and only the two low-rank matrices A and B need to be updated. When the model is output, BA is superimposed with the parameters of the pre-trained language model. During inference, h=Wx+BAx=(W+BA)x, so the matrix product B and the weight matrix W are added together as the new weight.

[0071] 4) Model training: Fine-tune the training based on the prepared data. Some model parameters are set as follows: the Lora rank parameter is 16 (the larger the rank, the more parameters and the slower the training), the training batch (epoch) is set to 30, the batch size is 32, the maximum number of steps executed in the training process (max_steps) is set to 5000, and the model is saved after training.

[0072] Step S103, when the task entity information of the current task content has not been collected, a progressive entity question of the user is generated, and in response to the user's reply operation, the question answer content of the progressive entity question is collected.

[0073] In this embodiment, when the task entity information of the current task content of the terminal has not been collected, a progressive entity question of the user is generated, and in response to the user's reply operation, the question answer content of the progressive entity question is collected. Among them, after each collection of the current task content, the terminal determines whether the task entity information has been collected, and if the collection is not completed, based on the task entity type corresponding to the task entity information that has not been collected, the user's progressive entity question is generated. The specific generation process will be described in detail later. Among them, the task entity type of the task entity information, for example, the life insurance reporting task needs to collect entity slots such as the name of the insured, the address of the insured, and the relationship between the reporter and the insured; the transfer task customer needs to provide the payee, transfer amount, etc.

[0074] Step S104, replace the user demand content with the question answer content, and return to execute the step of extracting the task entity information of the current task content based on the user demand content, until the task entity information of the current task content is collected, and then stop the iteration process.

[0075] In this embodiment, the terminal replaces the user request content with the question answer content, and returns to execute the step of extracting the task entity information of the current task content based on the user's request content, until the task entity information of the current task content is collected, and then stops the iterative process.

[0076] Step S105 , based on the task entity information obtained in each iteration, the user's question answering information is generated through the customer service strategy corresponding to the current task content.

[0077] In this embodiment, the terminal generates user question answering information based on the task entity information obtained in each iteration through the customer service strategy corresponding to the current task content. Among them, the customer service strategy includes the solution corresponding to the task entity data range corresponding to each task entity type of different task content. The specific generation process will be described in detail later.

[0078] Based on the above scheme, the user demand content of the user's question information and the current task content screened by the user are matched through the large model, so as to extract the task entity information related to the current task content in the user's question information, and when the entity information is not fully collected, the entity information is further collected through progressive question interaction to ensure that the entity information related to the current task content is fully collected. Finally, the terminal generates the user's question answer information based on the collected task entity information. It avoids the component stacking multi-module architecture of the traditional multi-round dialogue robot, and realizes the adaptive multi-round dialogue robot and system structure based on the large model. It does not rely on traditional machine learning modules, but uses large models to complete all dialogue interactions, significantly improving the semantic understanding and dialogue logic coherence of the dialogue system, and improving the robot interaction effect. Thereby effectively improving the intelligent interactive answer effect to user questions.

[0079] Optionally, identifying the user demand content corresponding to the current question information includes: based on the current question information, identifying the question semantic content corresponding to the current question information through a semantic recognition network; identifying the user question intent corresponding to the question semantic content through a text intent recognition network, and using the user question intent as the user demand content corresponding to the current question information.

[0080] In this embodiment, the terminal identifies the semantic content of the question corresponding to the current question information through the semantic recognition network based on the current question information. Among them, the semantic recognition network is a trained ChatGLM3-6B (6.2 billion parameters) model. Then, the terminal identifies the user's question intention corresponding to the question semantic content through the text intent recognition network, and uses the user's question intention as the user's demand content corresponding to the current question information. Among them, the text intent recognition network is a trained ChatGLM3-6B (6.2 billion parameters) model.

[0081] Based on the above solution, after semantic recognition, intent analysis is performed to improve the accuracy of user intent analysis.

[0082] Optionally, identifying the correlation information between the user's user demand content and the user's current task content, including: based on the user's current task content, extracting each prompt word information of the current task content through a keyword extraction strategy; based on each prompt word information and the user's demand content, identifying the similarity value between the user's demand content and each prompt word information through a semantic similarity recognition network, and using the similarity value between the user's demand content and each prompt word information as the correlation information between the user's user demand content and the user's current task content.

[0083] In this embodiment, the terminal extracts each prompt word information of the current task content through a keyword extraction strategy based on the user's current task content. Among them, in the keyword extraction strategy, the keyword range corresponding to the task content is stored, and then, based on the user's current task content, the terminal queries the keyword range corresponding to the current task content in the task database to obtain each prompt word information. Based on each prompt word information and the user's demand content, the similarity value between the user's demand content and each prompt word information is identified through a semantic similarity recognition network, and the similarity value between the user's demand content and each prompt word information is used as the correlation information between the user's demand content and the user's current task content. Among them, the semantic similarity recognition network is based on the trained ChatGLM3-6B (6.2 billion parameters) model for semantic similarity recognition.

[0084] Based on the above solution, the trained ChatGLM3-6B (6.2 billion parameters) model is used to analyze the correlation information between user demand content and the user's current task content, thereby improving the accuracy and efficiency of the analysis.

[0085] Optionally, when the correlation information satisfies a preset correlation condition, task entity information of the current task content is extracted based on the user's demand content, including: based on the user demand content and the similarity value between each prompt word information, identifying a first number of prompt word information corresponding to a similarity value greater than a similarity threshold, and calculating the correlation value of the user demand content based on the first number and the total number of all prompt word information; when the correlation value is greater than the correlation value threshold, determining that the correlation information satisfies the preset correlation condition, and performing entity information extraction processing on the user demand content through the entity extraction model corresponding to the current task content to obtain the task entity information of the current task content.

[0086] In this embodiment, the terminal identifies a first number of prompt word information corresponding to a similarity value greater than a similarity threshold based on the user's request content and the similarity value between each prompt word information, and uses the ratio between the first number and the total number of all prompt word information as the relevant value of the user's request content.

[0087] When the correlation value is greater than the correlation value threshold, the terminal determines that the correlation information meets the preset correlation condition, and performs entity information extraction processing on the user demand content through the entity extraction model corresponding to the current task content to obtain the task entity information of the current task content. When the correlation value is not greater than the correlation value threshold, the terminal returns to execute the step of identifying the user demand content corresponding to the current question information, thereby re-identifying the user's demand content to ensure the accuracy of identifying the user's demand content.

[0088] Based on the above scheme, by presetting correlation conditions, the standard for extracting user demand content is determined, thereby improving the accuracy of extracting user demand content and the accuracy of identifying the user's actual demand content.

[0089] Optionally, when the task entity information of the current task content has not been collected, a progressive entity question for the user is generated, including: based on the current task content, querying each task entity type corresponding to the current task content in the task database, and identifying each current task entity type contained in the task entity information of the current task content; based on each current task entity type, screening each missing task entity type without entity content in each task entity type, and randomly screening a target task entity type in each missing task entity type; in the question database, screening the question content of the target task entity type as the user's progressive entity question.

[0090] In this embodiment, the terminal queries each task entity type corresponding to the current task content in the task database based on the current task content, and identifies each current task entity type contained in the task entity information of the current task content. Then, based on each current task entity type, the terminal screens each missing task entity type without entity content in each task entity type, and randomly screens the target task entity type in each missing task entity type. Among them, randomly screening the target task entity type is to randomly screen one task entity type each time. Then, the terminal screens the question content of the target task entity type in the question database as the user's progressive entity question. Among them, the question database stores the question content corresponding to different task entity types, and the terminal queries the question content corresponding to each target task entity type by means of database query.

[0091] Based on the above scheme, interactive question generation is performed based on task entity types, which avoids the inefficiency and high cost of multi-module question generation, thereby improving the efficiency of question generation.

[0092] Optionally, based on the task entity information obtained in each iteration, the user's question answering information is generated through the customer service strategy corresponding to the current task content, including: based on the task entity information obtained in each iteration, identifying the entity content information of each task entity type; based on the entity content information of each task entity type, identifying the task entity data of each task entity type, and based on the task entity data of each task entity type, through the customer service strategy corresponding to the current task content, querying each problem answering solution corresponding to the current problem information; using all problem answering solutions as the user's problem answering information.

[0093] In this embodiment, the terminal identifies the entity content information of each task entity type based on the task entity information obtained in each iteration. Then, the terminal identifies the task entity data of each task entity type based on the entity content information of each task entity type, and based on the task entity data of each task entity type, queries the problem solving solutions corresponding to the current problem information through the customer service strategy corresponding to the current task content. Among them, the customer service strategy includes a list of problem solving solutions corresponding to each task entity data range of each task entity type. Then, based on the task entity content of each task entity type of the user, the terminal identifies the corresponding target task entity data range of each task entity type of the user in each task entity data range, and uses the list of problem solving solutions corresponding to each target task entity data range as the problem solving solution of the user. Finally, the terminal uses all problem solving solutions as the user's problem solving information.

[0094] Based on the above solution, the problem solving solution of the user is adapted in the problem solving solution through the task entity data of each task entity type, thereby improving the adaptation efficiency of the problem solving solution.

[0095] This application also provides an example of intelligent interactive answering of user questions, such as Figure 3 As shown, the specific processing process includes the following steps:

[0096] Step S301, obtaining the user's current question information and the user's current task content.

[0097] Step S302: Based on the current question information, the question semantic content corresponding to the current question information is identified through a semantic recognition network.

[0098] Step S303, identifying the user's question intention corresponding to the question semantic content through the text intention recognition network, and using the user's question intention as the user's demand content corresponding to the current question information.

[0099] Step S304: Based on the current task content of the user, the prompt word information of the current task content is extracted through the keyword extraction strategy.

[0100] Step S305, based on each prompt word information and the user's demand content, a semantic similarity recognition network is used to identify the similarity value between the user's demand content and each prompt word information, and the similarity value between the user's demand content and each prompt word information is used as the correlation information between the user's demand content and the user's current task content.

[0101] Step S306, based on the similarity values ​​between the user's request content and each prompt word information, identify the first number of prompt word information corresponding to the similarity value greater than the similarity threshold, and calculate the relevance value of the user's request content based on the first number and the total number of all prompt word information.

[0102] Step S307, when the correlation value is greater than the correlation value threshold, it is determined that the correlation information meets the preset correlation condition, and the entity information extraction process is performed on the user demand content through the entity extraction model corresponding to the current task content to obtain the task entity information of the current task content.

[0103] Step S308, based on the current task content, query the task database for each task entity type corresponding to the current task content, and identify each current task entity type contained in the task entity information of the current task content.

[0104] Step S309 , based on each current task entity type, in each task entity type, each missing task entity type without entity content is screened, and in each missing task entity type, a target task entity type is randomly screened.

[0105] Step S310, in the question database, filter the question content of the target task entity type as the user's progressive entity question.

[0106] Step S311, in response to the user's reply operation, collect the answer content of the progressive entity question.

[0107] Step S312, replace the user request content with the question answer content, and return to execute the step of extracting the task entity information of the current task content based on the user's request content, until the task entity information of the current task content is collected, and then stop the iterative process.

[0108] Step S313: Based on the task entity information obtained in each iteration, entity content information of each task entity type is identified.

[0109] Step S314, based on the entity content information of each task entity type, identify the task entity data of each task entity type, and based on the task entity data of each task entity type, query the problem solution corresponding to the current problem information through the customer service strategy corresponding to the current task content.

[0110] Step S315, all problem-solving solutions are used as the user's problem-solving information.

[0111] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0112] Based on the same inventive concept, the embodiment of the present application also provides an intelligent interactive answering device for user questions for implementing the intelligent interactive answering method for user questions involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the embodiments of the intelligent interactive answering device for one or more user questions provided below can refer to the limitations of the intelligent interactive answering method for user questions above, and will not be repeated here.

[0113] In an exemplary embodiment, Figure 4 As shown, a device for intelligent interactive answering of user questions is provided, comprising: an acquisition module 410, an extraction module 420, a collection module 430, an iteration module 440 and a generation module 450, wherein:

[0114] The acquisition module 410 is used to acquire the user's current question information and the user's current task content, and identify the user's demand content corresponding to the current question information;

[0115] The extraction module 420 is used to identify the correlation information between the user's user demand content and the user's current task content, and extract the task entity information of the current task content based on the user's user demand content when the correlation information meets a preset correlation condition;

[0116] The collection module 430 is used to generate a progressive entity question of the user when the task entity information of the current task content has not been collected, and collect the question answer content of the progressive entity question in response to the user's reply operation;

[0117] Iteration module 440, used to replace the user demand content with the question answer content, and return to execute the step of extracting the task entity information of the current task content based on the user demand content, until the task entity information of the current task content is collected, and then stop the iteration process;

[0118] The generation module 450 is used to generate the user's question answering information based on the task entity information obtained in each iteration and through the customer service strategy corresponding to the current task content.

[0119] Optionally, the acquisition module 410 is specifically configured to:

[0120] Based on the current question information, identifying the question semantic content corresponding to the current question information through a semantic recognition network;

[0121] The text intention recognition network is used to identify the user question intention corresponding to the question semantic content, and the user question intention is used as the user demand content corresponding to the current question information.

[0122] Optionally, the extraction module 420 is specifically used to:

[0123] Based on the current task content of the user, extracting each prompt word information of the current task content through a keyword extraction strategy;

[0124] Based on the prompt word information and the user demand content, the similarity value between the user demand content and the prompt word information is identified through a semantic similarity recognition network, and the similarity value between the user demand content and the prompt word information is used as the correlation information between the user demand content of the user and the user's current task content.

[0125] Optionally, the extraction module 420 is specifically used to:

[0126] Based on the similarity values ​​between the user demand content and each of the prompt word information, identifying a first number of prompt word information corresponding to a similarity value greater than a similarity threshold, and calculating a relevance value of the user demand content based on the first number and the total number of all prompt word information;

[0127] When the correlation value is greater than the correlation value threshold, it is determined that the correlation information meets the preset correlation condition, and the entity information extraction process is performed on the user demand content through the entity extraction model corresponding to the current task content to obtain the task entity information of the current task content.

[0128] Optionally, the acquisition module 430 is specifically used to:

[0129] Based on the current task content, querying each task entity type corresponding to the current task content in a task database, and identifying each current task entity type contained in the task entity information of the current task content;

[0130] Based on each of the current task entity types, in each of the task entity types, each missing task entity type without entity content is screened, and in each of the missing task entity types, a target task entity type is randomly screened;

[0131] In the question database, the question content of the target task entity type is screened as the progressive entity question of the user.

[0132] Optionally, the generating module 450 is specifically configured to:

[0133] Based on the task entity information obtained in each iteration, identifying entity content information of each task entity type;

[0134] Based on the entity content information of each task entity type, identifying the task entity data of each task entity type, and based on the task entity data of each task entity type, querying each problem solution corresponding to the current problem information through the customer service strategy corresponding to the current task content;

[0135] All problem-solving solutions are regarded as the problem-solving information of the user.

[0136] Each module in the above-mentioned intelligent interactive answering device for user questions can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0137] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 5As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be realized through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, an intelligent interactive solution method for user problems is realized. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0138] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0139] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps of a method for intelligent interactive answering of user questions when executing the computer program.

[0140] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method for intelligent interactive answering of user questions are implemented.

[0141] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the steps of the method for intelligent interactive answering of user questions.

[0142] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0143] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0144] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0145] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. An intelligent interactive solution method for user questions, characterized in that: The method comprises: Obtaining the user's current question information and the user's current task content, and identifying the user's demand content corresponding to the current question information; Identify the correlation information between the user's user demand content and the user's current task content, and extract the task entity information of the current task content based on the user's user demand content when the correlation information meets a preset correlation condition; When the task entity information of the current task content has not been collected, generating a progressive entity question for the user, and collecting the question answer content of the progressive entity question in response to the user's reply operation; The question answer content replaces the user demand content, and returns to the step of extracting the task entity information of the current task content based on the user demand content, until the task entity information of the current task content is collected, and then stops the iteration process; Based on the task entity information obtained in each iteration, the user's question answering information is generated through the customer service strategy corresponding to the current task content.

2. The method according to claim 1, characterized in that The identifying the user demand content corresponding to the current question information includes: Based on the current question information, identifying the question semantic content corresponding to the current question information through a semantic recognition network; The text intention recognition network is used to identify the user question intention corresponding to the question semantic content, and the user question intention is used as the user demand content corresponding to the current question information.

3. The method according to claim 1, characterized in that The identifying of the correlation information between the user's demand content and the user's current task content includes: Based on the current task content of the user, extracting each prompt word information of the current task content through a keyword extraction strategy; Based on the prompt word information and the user demand content, the similarity value between the user demand content and the prompt word information is identified through a semantic similarity recognition network, and the similarity value between the user demand content and the prompt word information is used as the correlation information between the user demand content of the user and the user's current task content.

4. The method according to claim 3, characterized in that: When the correlation information satisfies a preset correlation condition, extracting the task entity information of the current task content based on the user's user demand content includes: Based on the similarity values ​​between the user demand content and each of the prompt word information, identifying a first number of prompt word information corresponding to a similarity value greater than a similarity threshold, and calculating a relevance value of the user demand content based on the first number and the total number of all prompt word information; When the correlation value is greater than the correlation value threshold, it is determined that the correlation information meets the preset correlation condition, and the entity information extraction process is performed on the user demand content through the entity extraction model corresponding to the current task content to obtain the task entity information of the current task content.

5. The method according to claim 1, characterized in that: When the task entity information of the current task content has not been collected, generating the user's progressive entity question includes: Based on the current task content, querying each task entity type corresponding to the current task content in a task database, and identifying each current task entity type contained in the task entity information of the current task content; Based on each of the current task entity types, in each of the task entity types, each missing task entity type without entity content is screened, and in each of the missing task entity types, a target task entity type is randomly screened; In the question database, the question content of the target task entity type is screened as the progressive entity question of the user.

6. The method according to claim 1, characterized in that The task entity information obtained in each iteration is used to generate the user's question answering information through the customer service strategy corresponding to the current task content, including: Based on the task entity information obtained in each iteration, identifying entity content information of each task entity type; Based on the entity content information of each task entity type, identifying the task entity data of each task entity type, and based on the task entity data of each task entity type, querying each problem solution corresponding to the current problem information through the customer service strategy corresponding to the current task content; All problem-solving solutions are regarded as the problem-solving information of the user.

7. An intelligent interactive answering device for user questions, characterized in that: The device comprises: An acquisition module is used to acquire the user's current question information and the user's current task content, and identify the user's demand content corresponding to the current question information; An extraction module, used to identify the correlation information between the user's user demand content and the user's current task content, and extract the task entity information of the current task content based on the user's user demand content when the correlation information meets a preset correlation condition; A collection module, configured to generate a progressive entity question for the user when the task entity information of the current task content has not been collected, and collect the question answer content of the progressive entity question in response to the user's reply operation; An iteration module, used for replacing the user demand content with the question answer content, and returning to execute the step of extracting the task entity information of the current task content based on the user demand content, until the task entity information of the current task content is collected and the iteration process is stopped; A generation module is used to generate the user's question answering information based on the task entity information obtained in each iteration and through the customer service strategy corresponding to the current task content.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images