Question answer feedback method and device based on human-computer interaction, equipment and medium

Abstract

The invention relates to the technical field of artificial intelligence, can be applied to business system platforms of financial science and technology, medical health and the like, and discloses a man-machine interaction-based question answer feedback method, device, equipment and medium, the method comprises the following steps: obtaining a question help request of a target user, obtaining a question frequency, and if the question frequency is less than a preset frequency threshold, sending a question answer to the target user; if yes, intention recognition is conducted on the question help-seeking request to obtain question key information, and if yes, context reloading is conducted on an obtained historical question record according to the question help-seeking request to obtain the question key information, statement splitting is conducted on the question key information to obtain a key task; and searching related information of the key task from the obtained external professional knowledge base, generating a user preference behavior according to the historical question record, generating a question answer of the question help request according to the related information, and sending the question answer to a preset target user side. According to the invention, the efficiency of answering various questions for the user is improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method, device, equipment and medium for question answer feedback based on human-computer interaction. Background Art

[0002] Robotic assistants, especially AI-based conversational systems, have become essential tools for solving problems and providing answers. Leveraging natural language processing technology, these intelligent assistants can understand user questions, analyze context, quickly extract information from large amounts of data, and provide clear and concise responses. They are widely used in fields such as healthcare and fintech. However, traditional AI-based robot assistants struggle to effectively access user history to answer subsequent questions, requiring the call to a human customer service representative, resulting in inefficient problem resolution.

[0003] In the field of healthcare, AI-based robot assistants can provide patients with timely health advice, disease prevention recommendations, and drug information. Patients can use smart assistants to ask about symptoms, obtain preliminary health assessments, or consult about treatments for common diseases. However, when patients ask whether some symptoms are serious and need to make judgments based on their personal medical history and medication use, it is difficult for robot assistants to effectively access historical records of patients' medical history, medication use, and other historical questions, and they are unable to understand the patients' specific needs, resulting in lower efficiency in solving user problems.

[0004] In the field of financial technology, robot assistants can help users handle daily banking transactions, financial planning, and investment advice, etc., and query account balances, transfers, and investment portfolio performance through smart assistants. However, when users need robot assistants to provide financial advice based on their personal financial management situation, robot assistants cannot effectively call up historical question records such as historical investments, and need to transfer to manual customer service, resulting in lower efficiency in solving user problems.

[0005] In summary, existing bot Q&A applications often combine human interaction with intelligent robots. When users encounter complex or unsolvable issues while interacting with intelligent robots, they must be transferred to human customer service. Frequent page switching and multiple operations can lead to a poor user experience, and for unfamiliar users, the process can be cumbersome and unintuitive. Furthermore, intelligent bot systems often lack long-term memory and are unable to deeply understand users' past questions and preferences. This forces users to re-describe their situation each time they communicate, increasing duplication of effort and reducing the efficiency of resolving user issues.

[0006] Therefore, the current technology has the problem of low efficiency in answering various questions for users. Summary of the Invention

[0007] The present invention provides a method, device, equipment and medium for question answer feedback based on human-computer interaction, the main purpose of which is to solve the problem of low efficiency in answering various questions for users.

[0008] In a first aspect, to achieve the above-mentioned objectives, the present invention provides a method for providing feedback on a question answer based on human-computer interaction, comprising:

[0009] Obtaining a target user's problem assistance request, and obtaining the number of questions asked based on the problem assistance request;

[0010] If the number of questions asked is less than a preset threshold, the intention of the question help request is identified to obtain the key information of the target user's question;

[0011] If the number of questions asked is greater than or equal to a preset threshold, obtaining the historical question record of the target user, and reloading the context of the historical question record according to the question help request to obtain the key information of the question asked by the target user;

[0012] Decompose the key information of the question into sentences to obtain key tasks;

[0013] Obtaining an external professional knowledge base, and retrieving relevant information of the key task from the external professional knowledge base;

[0014] generating user preference behaviors of the target user according to the historical question records;

[0015] An answer to the question request for help is generated according to the user preference behavior and the relevant information, and the answer is sent to a preset target user terminal.

[0016] In a second aspect, the present invention further provides a question answer feedback device based on human-computer interaction, comprising:

[0017] A data acquisition module is used to obtain a target user's problem-solving request and obtain the number of questions asked based on the problem-solving request;

[0018] An intention recognition module is used to identify the intention of the question request if the number of questions asked is less than a preset number threshold, and obtain key information of the question asked by the target user;

[0019] a record reloading module configured to obtain the historical question records of the target user if the number of questions asked is greater than or equal to a preset number threshold, and to contextually reload the historical question records according to the question assistance request to obtain key information of the question asked by the target user;

[0020] A sentence decomposition module is used to decompose the key information of the question into sentences to obtain key tasks;

[0021] An information retrieval module is used to obtain an external professional knowledge base and retrieve relevant information of the key task from the external professional knowledge base;

[0022] A preference analysis module, configured to generate the user preference behavior of the target user based on the historical question records;

[0023] The answer generation module is used to generate an answer to the question help request based on the user preference behavior and the relevant information, and send the answer to a preset target user terminal.

[0024] In a third aspect, the present invention further provides an electronic device, comprising:

[0025] at least one processor; and,

[0026] a memory communicatively connected to the at least one processor; wherein,

[0027] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the above-mentioned question answer feedback method based on human-computer interaction.

[0028] In a fourth aspect, the present invention also provides a computer-readable storage medium, in which at least one computer program is stored. The at least one computer program is executed by a processor in an electronic device to implement the above-mentioned method for question answer feedback based on human-computer interaction.

[0029] The present invention obtains a problem help request from a target user, and obtains the number of questions asked based on the problem help request. If the number of questions asked is less than a preset number threshold, the problem help request is subjected to intention recognition to obtain the target user's question key information. If the number of questions asked is greater than or equal to the preset number threshold, the target user's historical question record is obtained, and the historical question record is context-reloaded based on the problem help request to obtain the target user's question key information. The question key information is sentence-decomposed to obtain key tasks. By sentence-decomposing the question key information, word segmentation and tagging, semantic unit division and mapping to key tasks are gradually completed, which can effectively improve the system's understanding of the user's complex questions. With processing capabilities, through the mapping of semantic units and key tasks, it can more accurately match service logic and realize multi-task parallel processing or sequential scheduling, thereby improving the response efficiency of the robot assistant in multi-round dialogues and multi-intention scenarios, obtain external professional knowledge base, retrieve relevant information of the key tasks from the external professional knowledge base, and split the huge knowledge base into manageable small fragments by slicing professional texts, making retrieval more efficient, generate the user preference behavior of the target user according to the historical question record, generate the question answer of the question help request according to the user preference behavior and the relevant information, and send the question answer to the preset target user end, which can effectively improve the efficiency of answering various questions for users. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0031] Figure 1 A schematic diagram of an application environment of a method for providing feedback on a question answer based on human-computer interaction in one embodiment of the present invention;

[0032] Figure 2 A flowchart of a method for providing feedback on a question answer based on human-computer interaction according to an embodiment of the present invention is provided;

[0033] Figure 3 A schematic flow chart of an information retrieval process in a question answer feedback method based on human-computer interaction provided by one embodiment of the present invention;

[0034] Figure 4 A schematic diagram of a module of a question answer feedback device based on human-computer interaction provided by one embodiment of the present invention;

[0035] Figure 5A schematic structural diagram of an electronic device for implementing a method for providing feedback on a question answer based on human-computer interaction according to an embodiment of the present invention;

[0036] Figure 6 Another structural diagram of an electronic device implementing a question answer feedback method based on human-computer interaction provided by an embodiment of the present invention.

[0037] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0038] In order to enable those skilled in the art to better understand the technical solutions of the present disclosure, and to fully understand and implement how the present disclosure applies technical means to solve technical problems and achieve the corresponding technical effects, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, not all of the embodiments. The embodiments of the present disclosure and the various features in the embodiments can be combined with each other without conflict, and the technical solutions formed are all within the scope of protection of the present disclosure. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of the present disclosure.

[0039] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, apparatus, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0040] An embodiment of the present application provides a method for feedback on questions and answers based on human-computer interaction, and the execution subject of the method for feedback on questions and answers based on human-computer interaction includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the device provided by the embodiment of the present application. In other words, the method for feedback on questions and answers based on human-computer interaction can be executed by software or hardware installed on a terminal device or a server device. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be an independent server, or it can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0041] The embodiment of the present invention provides a method for providing a feedback on a question answer based on human-computer interaction, which can be applied in the following situations: Figure 1 application environment. Among them, the client communicates with the server through the network. The server can obtain the problem help request of the target user through the client, and obtain the number of questions asked according to the problem help request. If the number of questions asked is less than the preset number threshold, the problem help request is intentionally identified to obtain the key information of the target user's question. If the number of questions asked is greater than or equal to the preset number threshold, the historical question record of the target user is obtained, and the historical question record is context-reloaded according to the problem help request to obtain the key information of the target user's question. The key information of the question is sentence-decomposed to obtain the key task. By sentence-decomposing the key information of the question, word segmentation and tagging, semantic unit division and mapping to key tasks are gradually completed, which can effectively improve the system's ability to handle complex questions of users. The ability to understand and process questions, through the mapping of semantic units and key tasks, can more accurately match the service logic, realize multi-task parallel processing or sequential scheduling, thereby improving the response efficiency of the robot assistant in multi-round dialogues and multi-intention scenarios, obtain external professional knowledge base, retrieve the relevant information of the key tasks from the external professional knowledge base, and by slicing professional texts, it can split the huge knowledge base into manageable small fragments, making the retrieval more efficient, generate the user preference behavior of the target user based on the historical question records, and generate the answer to the question request based on the user preference behavior and the relevant information, which can effectively improve the efficiency of answering various questions for users, and finally output the answer to the question and feed it back to the client. Among them, the client can be but not limited to various personal computers, laptops, smart phones, tablets and portable wearable devices. The server can be implemented with an independent server or a server cluster composed of multiple servers. The present invention is described in detail below through specific embodiments.

[0042] The following explains the present invention's description. The present invention introduces RAG technology, which enables more efficient integration with external professional knowledge bases to retrieve relevant information and generate accurate answers. RAG technology combines the retrieval and generation processes. First, using the Large Language Model (LLM) to retrieve documents or information relevant to the key task from external professional knowledge bases, it then combines this information with the current user question to generate a richer and more precise answer, improving the accuracy of the answer.

[0043] Reference Figure 2 FIG. 1 is a flow chart of a method for providing feedback on a question answer based on human-computer interaction according to an embodiment of the present invention. In this embodiment, the method for providing feedback on a question answer based on human-computer interaction includes:

[0044] S1. Obtain a problem help request from a target user, and obtain the number of questions asked according to the problem help request.

[0045] In this embodiment of the present invention, obtaining the target user's question assistance requests refers to collecting the questions or requests raised by the user when using the robot assistant. For example, the user may ask questions about health insurance, car insurance, life insurance, etc. These requests can be tracked and recorded in a log or database. A database query statement (such as SQL) is used to query and count the number of questions raised by the target user within a specified time period.

[0046] In specific healthcare scenarios, medical insurance claims often require a large amount of documents and evidence. Robot assistants can help users clearly understand the information and steps required for claims, and can also obtain a history of questions asked on the robot assistant.

[0047] In specific FinTech scenarios, users can interact directly with robot assistants through FinTech platforms to complete the entire process from selecting insurance products to paying premiums. The robot assistants will automate all tedious steps and provide users with a convenient insurance purchasing experience. At the same time, users can also obtain a history of questions asked on the robot assistants.

[0048] By collecting and tracking the target users' question records when using robot assistants, insurance companies or service providers can better understand user needs and behaviors. Not only can they accurately count the number of questions users ask in a specific time period, but they can also analyze users' attention to different insurance products, thereby providing personalized services and product recommendations, and improving customer satisfaction and service quality.

[0049] S2. Determine whether the number of questions asked is greater than or equal to a preset number threshold.

[0050] If the number of questions asked is less than the preset number threshold, then S3, perform intent recognition on the question help request to obtain key information of the target user's question.

[0051] In this embodiment of the present invention, by determining whether the number of questions asked by a target user reaches a preset threshold, it is possible to effectively screen out active users or users with high attention. If the number of questions asked is less than the preset threshold, it indicates that the target user's needs may be relatively basic or of limited interest. Therefore, further intent identification of the target user's questions is required. That is, by analyzing the user's specific requests and extracting their potential needs and intentions, this helps to more accurately understand the user's needs, provide more personalized and targeted services, and enhance the user experience.

[0052] In healthcare scenarios, such as mental health consultations, patients may frequently ask questions through online platforms, indicating emotional or psychological instability. By determining whether the number of questions exceeds or equals a threshold, the user is considered to be in need of additional mental health support. When this threshold is reached, a reminder can be triggered, prompting professional counselors to intervene to prevent the user's condition from worsening.

[0053] In specific FinTech scenarios, customers frequently ask questions about certain financial products such as loans, insurance, and investment products. We determine whether the number of questions asked is greater than or equal to a threshold, and whether the customer's interest in the product has increased. We then recommend related products or services to customers to improve customer experience and conversion rates.

[0054] In an embodiment of the present invention, the performing of intent recognition on the question-asking request to obtain key information of the target user's question includes:

[0055] Obtaining a request text of the problem assistance request, performing named entity recognition on the request text, and obtaining key entities;

[0056] Performing dependency syntax analysis on the request text to obtain text dependency relations;

[0057] Constructing a semantic vector of the request text according to the key entities and the text dependency;

[0058] Performing intent classification and recognition on the semantic vector to obtain an intent category;

[0059] Extracting key parameter information from the semantic vector;

[0060] Generate the target user's question key information according to the intention category and the key parameter information.

[0061] In detail, the request text is extracted from the request initiated by the user through the robot assistant. The request text usually contains insurance-related questions that the user wants to know, such as "I want to inquire about the claims process of auto insurance" or "Help me find medical insurance suitable for the elderly". The named entity recognition algorithm in natural language processing technology is used to process the request text to identify and extract key entities with semantic meaning, such as insurance type (such as "auto insurance", "medical insurance"), target population (such as "elderly"), operation actions (such as "query", "find"), etc., to obtain key entities.

[0062] Based on linguistic rules and statistical learning methods, the grammatical dependencies between words in a text are identified. This analysis determines the role of each word in a sentence and their dependencies, such as subject-predicate, verb-object, attributive-adverbial-complement, and so on. For example, in the request "Help me check the car insurance claim process," it can be identified that "help" is the main verb, "I" is the agent, "check" is the supplementary action, and "car insurance claim process" is the specific object of the query, thus identifying the text dependencies.

[0063] Each word in the request text is converted into a corresponding word vector. Combined with the key entity information extracted by named entity recognition, this process assigns higher weights to key entities, highlighting their importance in the overall semantics. Furthermore, the word dependency structure derived from dependency syntactic analysis is incorporated into the vector construction process. By guiding the encoding of structural relationships between words, this strengthens the structural understanding of semantics. Ultimately, the word vectors, entity features, and dependency structure information are integrated to produce the semantic vector for the request text.

[0064] Common insurance intentions include but are not limited to: policy inquiry, policy change, claim progress, claim inquiry, material consultation, insurance type recommendation, risk assessment, etc. The semantic vector of the request text is matched with the pre-defined intent categories, and the similarity between the semantic vector and each intent category is calculated to ultimately determine the most matching intent category.

[0065] By analyzing the word vectors in the semantic vectors, important information in the text is identified. The key parameter information in the semantic vectors includes: insurance type name, time information, operation object, operation behavior, etc. Based on the identified intent category and extracted key parameters, the system will generate an integrated question key information. For example, a user may ask "I need to know the coverage and compensation standards of health insurance." The system will extract "health insurance" (insurance type), "coverage" (key parameter), and "compensation standard" (key parameter) as the user's question key information.

[0066] Named entity recognition helps the system identify key information related to insurance. Dependency syntactic analysis further reveals the grammatical structure within the sentence, ensuring understanding of contextual relationships. Semantic vectors are constructed for intent classification and identification, which can effectively determine the user's true needs. Finally, key parameter information is extracted, enabling the system to provide users with personalized and accurate answers, optimize user experience and improve service quality.

[0067] If the number of questions asked is greater than or equal to the preset number threshold, then S4, obtain the historical question record of the target user, perform context reloading on the historical question record according to the question help request, and obtain the key information of the question asked by the target user.

[0068] In an embodiment of the present invention, when the system detects that the number of questions a user has asked reaches or exceeds a preset threshold, it will re-understand and analyze the user's needs based on the user's previous questions to better answer the current question. This ensures that the system takes into account the user's past context and intentions when answering questions, thereby providing more accurate and coherent answers and improving the user experience. Historical question records include: question text, timestamp, conversation turn number, reply status, conversation topic tags, and other data.

[0069] In healthcare scenarios, patients may frequently ask questions related to health conditions, disease prevention, medication use, treatment plans, etc. Based on the patient's question history, contextual reloading technology can be used to provide users with more accurate and personalized answers.

[0070] In specific fintech scenarios, based on the user's past loan inquiry records and financial status, the system can use contextual overloading technology to provide users with accurate information such as loan amounts and interest rates, and even automate the loan approval process.

[0071] In an embodiment of the present invention, context reloading the historical question record according to the question assistance request to obtain key information of the target user's question includes:

[0072] Extract keywords from the historical question records to obtain key question words;

[0073] Generate a contextual semantic profile based on the key question words;

[0074] Context-reloading the historical question record according to the context semantic portrait and the question help request to obtain a question semantic frame of the question help request;

[0075] Performing element parsing on the question semantic frame to obtain initial semantic elements;

[0076] Screening the initial semantic elements to obtain key semantic elements;

[0077] The key information of the target user's question is generated according to the key semantic elements.

[0078] Specifically, key question words represent the core content of the user's question, helping the system better understand the user's intent and providing a basis for subsequent Q&A and service. Key question words such as "insurance type," "claim," and "insurance amount" are identified through natural language processing technology.

[0079] All semantic information, including key question words and their contextual relationships, is integrated into a complete contextual semantic profile. This contextual semantic profile not only reflects the user's immediate needs but also potentially reflects underlying background information and preferences. For example, if a user frequently asks a certain type of insurance question, the system can identify and update the user's preferences, thereby providing more personalized recommendations.

[0080] By combining historical query records with the current contextual semantic profile, the system analyzes the parts of historical queries that are relevant to the current request and reloads this information. For example, if a user previously inquired about a certain type of insurance and now mentions similar insurance, the system can reload and compare historical data to update the user's need profile, ensuring a more accurate understanding of the current request.

[0081] The reloaded contextual information constructs a question semantic frame. This semantic frame is a high-level abstraction of the user's current question, encompassing the core intent of the user's question, key information (such as insurance type, coverage amount, and claim requirements), and background data related to previous questions. This semantic frame helps the robot assistant more accurately identify user needs and provide personalized and precise answers or suggestions.

[0082] Perform structured analysis of the various semantic elements within the question semantic frame, identifying key question types, intended verbs, related entities (such as insurance type, target population, time range, etc.), and contextual information. Semantic elements are categorized and weighted based on their roles and relationships within the question semantic frame, selecting the elements that best represent the user's current needs and combining them into clearly structured and semantically explicit key information for the question.

[0083] By contextually reloading historical question records for problem assistance requests, the semantic relevance of users' continuous conversations can be fully explored. Key question words from historical question records are first extracted to construct a contextual semantic profile that reflects the user's long-term focus and semantic preferences. Then, combined with the current request, the historical semantic content is dynamically reconstructed to generate a structured question semantic frame. The key question information extracted from this semantic frame can more accurately reflect the user's true intentions and personalized needs, significantly improving the robot assistant's understanding and response accuracy in complex conversation scenarios, enabling smarter and more user-friendly insurance service recommendations.

[0084] S5. Decompose the key information of the question into sentences to obtain key tasks.

[0085] In an embodiment of the present invention, the questions raised by the user are broken down into smaller, more specific tasks or information units so that the system can understand and process each part more accurately. Through such decomposition, the system can process each task separately to ensure that the answers are clearer, complete and relevant.

[0086] In specific healthcare scenarios, patients often ask questions related to health conditions, disease prevention, medication use, treatment plans, etc. Based on the patient's question history, we can break down the question history into sentences to obtain the keywords asked by the patient, providing analytical data for providing more accurate diagnostic answers to the patient.

[0087] In specific financial technology scenarios, users will ask questions about loan records and financial status. By breaking down the user's loan question records and obtaining the keywords in the user's questions, we can provide analytical data for providing more accurate loan-related answers in the future.

[0088] In an embodiment of the present invention, the sentence decomposition of the key information of the question to obtain the key tasks includes:

[0089] Performing word segmentation and annotation on the key information of the question to obtain a number of question words;

[0090] Dividing the key information of the question into semantic units according to the question words to obtain a plurality of question semantic units;

[0091] The question semantic unit is mapped into several key tasks.

[0092] Specifically, the key information of the input question is standardized, including removing special characters, unifying full-width and half-width symbols, etc., and the continuous text is divided into several question words with independent meanings. Each word segmentation result is tagged with part of speech to identify its grammatical role in the sentence, such as verbs, nouns, time words, numerals, prepositions, etc. This step helps with subsequent semantic recognition and task classification.

[0093] Perform semantic clustering and structural division on the question words, identify the semantic associations and functional boundaries between words, and divide the key information of the question into several semantic units. Each semantic unit corresponds to a relatively independent semantic fragment. According to the preset semantic unit mapping rules, perform semantic understanding and functional matching on each semantic unit, judge the task intention corresponding to the semantic unit, and thus generate the corresponding key tasks.

[0094] By breaking down the key information in the question into sentences, gradually completing word segmentation and tagging, dividing semantic units, and mapping them into key tasks, the system can effectively improve its ability to understand and handle complex user questions. Not only can it refine the multiple intentions expressed by the user, but it can also structure lengthy or ambiguous request statements into clear task units, enhancing the accuracy and operability of semantic processing. At the same time, through the mapping of semantic units to key tasks, the system can more accurately match service logic and achieve multi-task parallel processing or sequential scheduling, thereby improving the response efficiency and service intelligence level of the robot assistant in multi-round dialogue and multi-intention scenarios.

[0095] S6. Obtain an external professional knowledge base, and retrieve relevant information of the key task from the external professional knowledge base.

[0096] In this embodiment of the present invention, the introduction of RAG (Retrieval-Augmented Generation) technology enables more efficient integration with external professional knowledge bases, retrieving relevant information from them and generating accurate answers. RAG technology combines the retrieval and generation processes. First, it uses the Large Language Model (LLM) to obtain documents or information related to key tasks from external professional knowledge bases. Then, it combines this information with the current user question to generate a richer and more precise answer, improving the accuracy of the answer.

[0097] External professional knowledge bases include: commercial insurance terms database, industry policy and regulations database, historical claims case database, expert question and answer knowledge graph, medical terminology or property valuation knowledge system, etc.

[0098] In specific medical and health scenarios, doctors can use external knowledge bases such as PubMed and Cochrane Library to retrieve the latest medical research results related to keywords in patient user's question history, helping doctors make more accurate treatment decisions.

[0099] In specific FinTech scenarios, external financial market data, such as stock prices and bond market data, can be used to analyze market movements and trends. Answering users' questions based on keywords in their loan inquiry records can help investment managers or financial analysts make more forward-looking decisions.

[0100] Figure 3 A flowchart of an information retrieval process in a question answer feedback method based on human-computer interaction provided by an embodiment of the present invention.

[0101] In the embodiment of the present invention, the retrieving relevant information of the key task from the external professional knowledge base includes:

[0102] Slicing the external professional knowledge base to obtain a plurality of professional texts;

[0103] Converting the professional text into a text vector;

[0104] Converting the key task into a task query vector;

[0105] Calculating the similarity between the text vector and the task query vector, and obtaining a number of similar knowledge segments based on the similarity;

[0106] The source documents of the similar knowledge fragments are obtained, and several of the knowledge fragments and the corresponding source documents are aggregated to obtain relevant information of the key task.

[0107] Specifically, text slicing operations on the content of external professional knowledge bases refer to dividing the entire library text into several professional text segments according to logical paragraphs, semantic boundaries or fixed lengths to ensure that each piece of content has semantic integrity.

[0108] Preprocess professional texts, including word segmentation and stop word removal, encode the text using a pre-trained language model (such as BERT, GPT, Word2Vec, etc.), convert the text into a fixed-length vector representation, and obtain the text vector and task query vector.

[0109] The similarity between the text vector and the task query vector is calculated. Based on the calculated similarity value, the text segment that is most similar to the task query can be selected. These text segments typically have high semantic relevance and can provide useful information for task execution. A highly similar text segment means that the content of the segment is very close to the task requirements in semantics, and therefore can be used as a reference or supplementary knowledge to help complete the task. The similarity calculation formula is as follows:

[0110]

[0111] Where A represents the text vector, B represents the task query vector, ||A|| represents the modulus of the text vector, and ||B|| represents the modulus of the task query vector.

[0112] After calculating the similarity between the text vector and the task query vector, the similarity value can be used to determine which knowledge fragments are most relevant to the task query. For each relevant knowledge fragment, the system will track the document from which the fragment comes. These documents are usually structured text data, articles or research reports, etc. Each relevant knowledge fragment is matched with the document from which it comes and summarized. Each knowledge fragment corresponds to its source and can be traced back to the original document. This aggregation process ensures that the obtained knowledge fragments are closely related to the task query, and also provides detailed source information to help users or systems understand the background and source of these knowledge fragments. The summarized knowledge fragments and source documents will be integrated into a set of relevant information to obtain relevant information for the key task.

[0113] By slicing professional text, the vast knowledge base can be broken down into manageable fragments, making retrieval more efficient. Next, by converting professional text and task queries into vector representations, semantic information can be accurately captured to ensure matching task requirements. Calculating the similarity between text vectors and task query vectors allows for rapid identification of task-related knowledge fragments, reducing the time and cost of information retrieval. Finally, these knowledge fragments and corresponding source documents are aggregated to provide detailed and reliable background information for the task, ensuring sufficient professional support when handling the task. This approach improves the accuracy of task execution and enhances the intelligence and responsiveness of the system.

[0114] S7: Generate user preference behavior of the target user based on the historical question records.

[0115] In an embodiment of the present invention, the user preference behavior of the target user is generated based on the historical question records, mainly by analyzing the user's past question content, frequency, topics of interest and interaction methods and other data to infer the user's interests, needs and habits, so as to provide more customized and accurate services and recommendations.

[0116] In specific medical and health scenarios, based on the patient's historical question records, we can analyze what kind of questions the patient prefers to ask. For example, if the patient has specific medication needs or long-term medication habits, we can analyze the medical history that the patient is most likely to have, and provide more accurate answers based on the medical history.

[0117] In specific FinTech scenarios, based on the user's historical question records, we can analyze what kind of questions the patient users prefer to ask. For example, if the user often consults about automated investment or financial management, we can analyze the user's preferred investment strategy and provide the user with more satisfactory answers later.

[0118] In an embodiment of the present invention, generating the user preference behavior of the target user based on the historical question record includes:

[0119] Extracting user behavior data from the historical question records;

[0120] Classifying the user behavior data to obtain user behavior labels;

[0121] Counting the user behavior tags, and calculating the tag weight of each user behavior tag based on the statistical number;

[0122] The user preference behavior of the target user is generated according to the tag weight.

[0123] In detail, user behavior data includes: historical question records of target users, interactive behavior characteristics, completed task tags, reading time, preferred topics, etc.

[0124] Based on the structured semantic elements, user behaviors are classified and labeled. User behavior labels include: high-frequency insurance type attention (such as "frequent attention to critical illness insurance"), common intention patterns (such as "repeated active consultation on claims process"), usage preference (such as "prefer text questions, tend to be manually transferred"), time period preference (such as "frequent questions at night", "concentrated questions before holidays"), service emotion tendency (such as "sensitive to price", "claims anxious"), etc.

[0125] Based on the frequency of user tags, we count the number of times each tag appears in user behavior, reflecting the tag's prevalence and importance. Furthermore, by considering the time interval and sequential position of tags, we can reveal whether users' preference for certain tags is persistent or sporadic, as well as the priority and influence of tags in user behavior. The TF-IDF strategy calculates a weighted score for each tag by weighing its frequency and distribution across the user population. High-weighted tags indicate a high degree of representativeness in user behavior, effectively reflecting users' interests and preferences, and ultimately identifying preferred user behavior.

[0126] Extracting user behavior data from historical question records helps comprehensively collect user interaction information. This behavior data is then categorized and labeled to clearly identify user behavioral characteristics. By counting the number of tags and calculating the weight of each tag, we can identify the topics and areas that users are most interested in, thereby tailoring more precise recommendations and services for users. This allows us to generate personalized user preference models based on real historical data, improving service relevance and user experience.

[0127] S8. Generate an answer to the question request based on the user preference behavior and the relevant information, and send the answer to a preset target user terminal.

[0128] In an embodiment of the present invention, based on the user's preference behavior and related information, the system identifies the insurance products, terms or claims that the user is most concerned about. The system retrieves relevant information from an external professional knowledge base, combines the user's preference behavior, and generates detailed answers that meet the user's needs. The answers are sent to the target user end through a robot assistant to ensure that users can quickly obtain the insurance information they need.

[0129] In specific medical and health scenarios, customized insurance services are provided to users based on their preferred behaviors and related information. For example, if a user has a history of chronic disease or has similar conditions in their family, the recommended insurance plan will specifically cover the treatment of related diseases, and provide regular physical examinations and drug subsidies.

[0130] In specific FinTech scenarios, customized insurance services are provided to users based on their preferences and related information. For example, a robot assistant will automatically remind you to renew your insurance based on the expiration date of your insurance and provide users with a simple renewal process.

[0131] In an embodiment of the present invention, generating an answer to the question request for help based on the user preference behavior and the relevant information, and sending the answer to a preset target user terminal, includes:

[0132] Obtaining professional terms in the key information of the question, parsing the professional terms according to the relevant information, and obtaining explanations of the professional terms;

[0133] Analyze the key information of the question based on the relevant information to obtain answer suggestions;

[0134] Filter and sort the professional term explanations and the answer suggestions according to the user's preference behavior to obtain the answer to the question request;

[0135] The answer to the question is sent to a preset target user terminal.

[0136] Specifically, by extracting professional terms from questions, we identify technical or domain terms that are crucial to understanding the user's questions, and then analyze these professional terms in detail based on relevant information or materials in the knowledge base to provide clear and accurate definitions and background.

[0137] By answering the key questions based on relevant information, the system can quickly and accurately generate suggested answers based on the key elements of the question. By analyzing the core content of the question, the system can extract relevant data, information, and professional explanations from the knowledge base or external resources and integrate them into a clear and organized answer.

[0138] Analyze all the answer suggestions generated by the robot assistant and extract the key information in the answers. Match the answer suggestions according to the user's behavioral tags. For example, if the user prefers low-risk insurance products, answers related to low-risk insurance will be displayed first. Based on the user's behavioral preferences and tag weights, the system filters out the answers that best match the user's needs. The filtered answer suggestions will be sorted by relevance to ensure that the answers that best meet the user's preferences are ranked first. For example, if the user has frequently searched for "medical insurance" recently, relevant medical insurance suggestions will be displayed first. The sorted answer suggestions are integrated into a complete answer, which may include multiple recommendations or detailed explanations to fully answer the user's question. The final sorted answer will be sent to the preset target user end through the system, such as the robot assistant's interface.

[0139] By parsing professional terms, we ensure that accurate knowledge answers are provided; then, we generate answer suggestions based on relevant information and user needs, and intelligently filter and sort them according to user preferences to ensure that the answers are more targeted and practical. This not only improves the accuracy of the answers, but also effectively enhances the user experience, because the answers are more in line with the user's interests and needs, and are delivered to the user end in a timely manner, saving users time in finding information and enhancing the convenience and satisfaction of the service.

[0140] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0141] like Figure 4 , which is a functional module diagram of a question answer feedback device based on human-computer interaction provided by one embodiment of the present invention.

[0142] In the embodiment of the present disclosure, a question answer feedback device based on human-computer interaction is provided, and the question answer feedback device based on human-computer interaction corresponds one to one with the question answer feedback method based on human-computer interaction in the above embodiment. Figure 4As shown, the human-computer interaction-based question answer feedback device 100 can be installed in an electronic device. According to the functions to be implemented, the human-computer interaction-based question answer feedback device 100 includes a data acquisition module 101, an intention recognition module 102, a record reloading module 103, a sentence decomposition module 104, an information retrieval module 105, a preference analysis module 106, and an answer generation module 107. The functional modules are described in detail as follows:

[0143] The data acquisition module 101 is used to obtain a target user's problem-solving request and obtain the number of questions asked based on the problem-solving request;

[0144] Intention recognition module 102, configured to perform intention recognition on the question request if the number of questions asked is less than a preset threshold, and obtain key information of the question asked by the target user;

[0145] The record reloading module 103 is configured to obtain the historical question record of the target user if the number of questions asked is greater than or equal to a preset number threshold, and contextually reload the historical question record according to the question help request to obtain key information of the question asked by the target user;

[0146] The sentence decomposition module 104 is used to decompose the key information of the question into key tasks;

[0147] An information retrieval module 105 is configured to obtain an external professional knowledge base and retrieve relevant information of the key task from the external professional knowledge base;

[0148] A preference analysis module 106 is configured to generate the user preference behavior of the target user based on the historical question records;

[0149] The answer generation module 107 is configured to generate an answer to the question request based on the user preference behavior and the relevant information, and send the answer to a preset target user terminal.

[0150] In one embodiment, when performing intent recognition on the question-asking request and obtaining key information of the target user's question, the intent recognition module 102 is configured to:

[0151] Obtaining a request text of the problem assistance request, performing named entity recognition on the request text, and obtaining key entities;

[0152] Performing dependency syntax analysis on the request text to obtain text dependency relations;

[0153] Constructing a semantic vector of the request text according to the key entities and the text dependency;

[0154] Performing intent classification and recognition on the semantic vector to obtain an intent category;

[0155] Extracting key parameter information from the semantic vector;

[0156] Generate the target user's question key information according to the intention category and the key parameter information.

[0157] In one embodiment, when the record reloading module 103 performs context reloading on the historical question record according to the question assistance request to obtain the key information of the question of the target user, it is configured to:

[0158] Extract keywords from the historical question records to obtain key question words;

[0159] Generate a contextual semantic profile based on the key question words;

[0160] Context-reloading the historical question record according to the context semantic portrait and the question help request to obtain a question semantic frame of the question help request;

[0161] Performing element parsing on the question semantic frame to obtain initial semantic elements;

[0162] Screening the initial semantic elements to obtain key semantic elements;

[0163] The key information of the target user's question is generated according to the key semantic elements.

[0164] In one embodiment, when the sentence decomposition module 104 decomposes the key information of the question to obtain the key task, it is used to:

[0165] Performing word segmentation and annotation on the key information of the question to obtain a number of question words;

[0166] Dividing the key information of the question into semantic units according to the question words to obtain a plurality of question semantic units;

[0167] The question semantic unit is mapped into several key tasks.

[0168] In one embodiment, when retrieving the relevant information of the key task from the external professional knowledge base, the information retrieval module 105 is configured to:

[0169] Slicing the external professional knowledge base to obtain a plurality of professional texts;

[0170] Converting the professional text into a text vector;

[0171] Converting the key task into a task query vector;

[0172] Calculating the similarity between the text vector and the task query vector, and obtaining a number of similar knowledge segments based on the similarity;

[0173] The source documents of the similar knowledge fragments are obtained, and several of the knowledge fragments and the corresponding source documents are aggregated to obtain relevant information of the key task.

[0174] In one embodiment, when generating the user preference behavior of the target user based on the historical question record, the preference analysis module 106 is configured to:

[0175] Extracting user behavior data from the historical question records;

[0176] Classifying the user behavior data to obtain user behavior labels;

[0177] Counting the user behavior tags, and calculating the tag weight of each user behavior tag based on the statistical number;

[0178] The user preference behavior of the target user is generated according to the tag weight.

[0179] In one embodiment, when the answer generation module 107 generates an answer to the question request for help based on the user preference behavior and the relevant information, and sends the answer to a preset target user terminal, it is configured to:

[0180] Obtaining professional terms in the key information of the question, parsing the professional terms according to the relevant information, and obtaining explanations of the professional terms;

[0181] Analyze the key information of the question based on the relevant information to obtain answer suggestions;

[0182] Filter and sort the professional term explanations and the answer suggestions according to the user's preference behavior to obtain the answer to the question request;

[0183] The answer to the question is sent to a preset target user terminal.

[0184] In the present invention, for a question answer feedback device based on human-computer interaction, first, the present invention obtains the problem help request of the target user, and obtains the number of questions asked according to the problem help request. If the number of questions asked is less than a preset number threshold, the intention of the problem help request is identified to obtain the key information of the target user's question. If the number of questions asked is greater than or equal to the preset number threshold, the historical question record of the target user is obtained, the context of the historical question record is reloaded according to the problem help request to obtain the key information of the target user's question, the key information of the question is sentence-decomposed to obtain the key task. By sentence-decomposing the key information of the question, word segmentation and tagging, semantic unit division and mapping to key tasks are gradually completed, which can effectively Improve the system's ability to understand and handle complex user questions. By mapping semantic units with key tasks, it can more accurately match service logic and achieve multi-task parallel processing or sequential scheduling, thereby improving the response efficiency of the robot assistant in multi-round dialogues and multi-intention scenarios, obtain an external professional knowledge base, and retrieve relevant information of the key tasks from the external professional knowledge base. By slicing professional texts, the huge knowledge base can be split into manageable small fragments, making retrieval more efficient. The user preference behavior of the target user is generated based on the historical question record, and the question answer to the question help request is generated based on the user preference behavior and the relevant information, and the question answer is sent to the preset target user terminal, which can effectively improve the efficiency of answering various questions for users. For the specific definition of a question answer feedback device based on human-computer interaction, please refer to the definition of a question answer feedback method based on human-computer interaction above, which will not be repeated here. The various modules in the above-mentioned question answer feedback device based on human-computer interaction can be implemented in whole or in part by software, hardware and their combination. The above modules may be embedded in or independent of the processor in the computer device in the form of hardware, or may be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0185] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 5As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external client via a network connection. When the computer program is executed by the processor, it realizes the functions or steps on the server side of a question answer feedback method based on human-computer interaction.

[0186] In one embodiment, a computer device is provided. The computer device may be a client, and its internal structure diagram may be as follows: Figure 6 As shown. The computer device includes a processor, memory, network interface, display screen and input device connected via a system bus. 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 computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it implements the functions or steps on the client side of a question answer feedback method based on human-computer interaction.

[0187] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are performed:

[0188] Obtaining a target user's problem assistance request, and obtaining the number of questions asked based on the problem assistance request;

[0189] If the number of questions asked is less than a preset threshold, the intention of the question help request is identified to obtain the key information of the target user's question;

[0190] If the number of questions asked is greater than or equal to a preset threshold, obtaining the historical question record of the target user, and reloading the context of the historical question record according to the question help request to obtain the key information of the question asked by the target user;

[0191] Decompose the key information of the question into sentences to obtain key tasks;

[0192] Obtaining an external professional knowledge base, and retrieving relevant information of the key task from the external professional knowledge base;

[0193] generating user preference behaviors of the target user according to the historical question records;

[0194] An answer to the question request for help is generated according to the user preference behavior and the relevant information, and the answer is sent to a preset target user terminal.

[0195] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and apparatuses can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the module division is merely a logical function division, and actual implementation may employ other division methods.

[0196] In addition, the functional modules in various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or hardware plus software functional modules.

[0197] Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the claims are intended to be embraced therein. Any reference to a figure in a claim should not be construed as limiting the claim to which it relates.

[0198] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0199] In some implementations of this embodiment, a computer-readable storage medium is provided, on which a computer program is stored, characterized in that when the computer program is executed by a processor, the steps of the method described in the above embodiment are implemented.

[0200] The readable storage medium of the present invention stores a computer program, which, when executed by a processor of an electronic device, can implement:

[0201] Obtaining a target user's problem assistance request, and obtaining the number of questions asked based on the problem assistance request;

[0202] If the number of questions asked is less than a preset threshold, the intention of the question help request is identified to obtain the key information of the target user's question;

[0203] If the number of questions asked is greater than or equal to a preset threshold, obtaining the historical question record of the target user, and reloading the context of the historical question record according to the question help request to obtain the key information of the question asked by the target user;

[0204] Decompose the key information of the question into sentences to obtain key tasks;

[0205] Obtaining an external professional knowledge base, and retrieving relevant information of the key task from the external professional knowledge base;

[0206] generating user preference behaviors of the target user according to the historical question records;

[0207] An answer to the question request for help is generated according to the user preference behavior and the relevant information, and the answer is sent to a preset target user terminal.

[0208] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can be found in the relevant descriptions of the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0209] The computer-readable storage medium may also store at least one computer-executable program / instruction, such as a computer-readable instruction. Computer-readable storage media include, but are not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory. Computer-readable storage media may include, for example, read-only memory (ROM), a hard disk, a flash memory, etc. For example, a non-transitory computer-readable storage medium may be connected to a computing device such as a computer, and then, when the computing device executes the computer-readable instructions stored on the computer-readable storage medium, the various methods described above may be performed.

[0210] In addition, the computer device may also include (but is not limited to) a data bus, an input / output (I / O) bus, a display, and input / output devices (eg, keyboard, mouse, speaker, etc.).

[0211] The processor can communicate with external devices via an I / O bus via a wired or wireless network.

[0212] In one embodiment, the at least one computer executable instruction may also be compiled into or constitute a software product / computer program product, wherein one or more computer executable instructions are executed by a processor to perform the various functions and / or method steps in the embodiments described in the present technology.

[0213] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. 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 memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0214] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0215] In the embodiments provided in the present disclosure, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions and operations of the devices, methods and computer program products according to multiple embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a portion of code, and the above-mentioned module, program segment or a portion of code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.

[0216] It should be noted that, in this disclosure, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element limited by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0217] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

[0218] It should be noted that if software tools or components other than those of our company appear in the embodiments of this application, they are only used for illustration and do not represent actual use.

Claims

1. A method for question answer feedback based on human-computer interaction, characterized in that: The method comprises: Obtaining a target user's problem assistance request, and obtaining the number of questions asked based on the problem assistance request; If the number of questions asked is less than a preset threshold, the intention of the question help request is identified to obtain the key information of the target user's question; If the number of questions asked is greater than or equal to a preset threshold, obtaining the historical question record of the target user, and reloading the context of the historical question record according to the question help request to obtain the key information of the question asked by the target user; Decompose the key information of the question into sentences to obtain key tasks; Obtaining an external professional knowledge base, and retrieving relevant information of the key task from the external professional knowledge base; generating user preference behaviors of the target user according to the historical question records; An answer to the question request for help is generated according to the user preference behavior and the relevant information, and the answer is sent to a preset target user terminal.

2. The method for question answer feedback based on human-computer interaction according to claim 1, characterized in that: The intention recognition of the question request is performed to obtain key information of the target user's question, including: Obtaining a request text of the problem assistance request, performing named entity recognition on the request text, and obtaining key entities; Performing dependency syntax analysis on the request text to obtain text dependency relations; Constructing a semantic vector of the request text according to the key entities and the text dependency; Performing intent classification and recognition on the semantic vector to obtain an intent category; Extracting key parameter information from the semantic vector; Generate the target user's question key information according to the intention category and the key parameter information.

3. The method for question answer feedback based on human-computer interaction according to claim 1, characterized in that: The context reloading of the historical question record according to the question assistance request to obtain key information of the target user's question includes: Extract keywords from the historical question records to obtain key question words; Generate a contextual semantic profile based on the key question words; Context-reloading the historical question record according to the context semantic portrait and the question help request to obtain a question semantic frame of the question help request; Performing element parsing on the question semantic frame to obtain initial semantic elements; Screening the initial semantic elements to obtain key semantic elements; The key information of the target user's question is generated according to the key semantic elements.

4. The method for question answer feedback based on human-computer interaction according to claim 1, characterized in that: The key information of the question is decomposed into sentences to obtain key tasks, including: Performing word segmentation and annotation on the key information of the question to obtain a number of question words; Dividing the key information of the question into semantic units according to the question words to obtain a plurality of question semantic units; The question semantic unit is mapped into several key tasks.

5. The method for question answer feedback based on human-computer interaction according to claim 1, characterized in that: The retrieving relevant information of the key task from the external professional knowledge base includes: Slicing the external professional knowledge base to obtain a plurality of professional texts; Converting the professional text into a text vector; Converting the key task into a task query vector; Calculating the similarity between the text vector and the task query vector, and obtaining a number of similar knowledge segments based on the similarity; The source documents of the similar knowledge fragments are obtained, and several of the knowledge fragments and the corresponding source documents are aggregated to obtain relevant information of the key task.

6. The method for question answer feedback based on human-computer interaction according to claim 1, characterized in that: Generating the user preference behavior of the target user according to the historical question record includes: Extracting user behavior data from the historical question records; Classifying the user behavior data to obtain user behavior labels; Counting the user behavior tags, and calculating the tag weight of each user behavior tag based on the statistical number; The user preference behavior of the target user is generated according to the tag weight.

7. The method for question answer feedback based on human-computer interaction according to claim 1, characterized in that: Generating an answer to the question request for help based on the user preference behavior and the relevant information, and sending the answer to a preset target user terminal, includes: Obtaining professional terms in the key information of the question, parsing the professional terms according to the relevant information, and obtaining explanations of the professional terms; Analyze the key information of the question based on the relevant information to obtain answer suggestions; Filter and sort the professional term explanations and the answer suggestions according to the user's preference behavior to obtain the answer to the question request; The answer to the question is sent to a preset target user terminal.

8. A question answer feedback device based on human-computer interaction, characterized in that: The device comprises: A data acquisition module is used to obtain a target user's problem-solving request and obtain the number of questions asked based on the problem-solving request; An intention recognition module is used to identify the intention of the question request if the number of questions asked is less than a preset number threshold, and obtain key information of the question asked by the target user; a record reloading module configured to obtain the historical question records of the target user if the number of questions asked is greater than or equal to a preset number threshold, and to contextually reload the historical question records according to the question assistance request to obtain key information of the question asked by the target user; A sentence decomposition module is used to decompose the key information of the question into sentences to obtain key tasks; An information retrieval module is used to obtain an external professional knowledge base and retrieve relevant information of the key task from the external professional knowledge base; A preference analysis module, configured to generate the user preference behavior of the target user based on the historical question records; The answer generation module is used to generate an answer to the question help request based on the user preference behavior and the relevant information, and send the answer to a preset target user terminal.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the question answer feedback method based on human-computer interaction as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for question answer feedback based on human-computer interaction as described in any one of claims 1 to 7 is implemented.

Images