Hotspot Q&A assistant system and hotspot Q&A implementation method based on hotspot Q&A assistant system

Abstract

This invention provides a hot topic question-and-answer assistant system and a method for implementing hot topic question-and-answer based on the hot topic question-and-answer assistant system. The method includes: parsing the original questions from users to extract the key information points of the original questions; constructing a thought chain based on all extracted key information points; integrating the constructed thought chain to obtain a question set; extracting the feature vectors of the question set; using vector similarity analysis to match the question set with the one with the highest similarity among all preset hot topic knowledge bases, which is used as the target hot topic knowledge base for question retrieval; retrieving a retrieval sub-result in the target hot topic knowledge base using a multi-way retrieval strategy according to the retrieval enhancement method; integrating all the retrieval sub-results obtained by the multi-way retrieval strategy for all questions contained in the question set to obtain a comprehensive question retrieval result; validating the comprehensive question retrieval result; and outputting the verified comprehensive question retrieval result.

Description

Technical Field

[0001] This invention relates to the field of large-scale question-and-answer assistant technology, and in particular to a hot topic question-and-answer assistant system and a hot topic question-and-answer implementation method based on the hot topic question-and-answer assistant system. Background Technology

[0002] Large language models, with the emergence of chatGPT, have become one of the most prominent research topics in the field of deep learning. Question-answering assistants, leveraging the capabilities of large models, are widely used. Hot topic question-answering assistants aim to quickly respond to and answer users' questions about current hot events, news, or popular topics. This requires the system to not only have strong language understanding capabilities but also the ability to acquire and understand the latest information in real-time or near real-time, and effectively integrate this information to generate accurate and timely answers.

[0003] In existing technologies, creating a hot topic question-answering assistant based on a large language model can be mainly achieved through the following methods:

[0004] (1) Retrieval-based question answering system: This type of system builds a large-scale knowledge base or index. When a user asks a question, the system retrieves the most relevant documents or paragraphs from the knowledge base and then extracts or generates the answer from them.

[0005] While this approach works well for handling common issues, it may fail to provide up-to-date information when dealing with trending topics due to the limited update speed of the knowledge base. Furthermore, since large models are trained on historical data, they may not be able to immediately acquire and understand the latest information for real-time or near-real-time trending events, thus limiting the timeliness and accuracy of the responses.

[0006] (2) Generative question answering system: This type of system utilizes the generative capabilities of pre-trained language models to directly generate answers based on questions.

[0007] This approach is flexible when dealing with open-domain problems, but the quality of the generated answers is highly dependent on the model's training data and parameter size, and may not provide accurate and up-to-date information when dealing with hot topics without real-time updated training data.

[0008] (3) Hybrid question answering system: It combines retrieval and generation strategies. First, relevant information is obtained through retrieval, and then the information is integrated using a generative model to generate the answer.

[0009] When dealing with trending issues, this method can obtain the latest information through real-time updated search sources, and then use the flexibility of generative models to integrate and generate answers, thereby providing more accurate and timely responses.

[0010] However, existing technologies that create hot topic question-answering assistants based on large language models often treat the large model as a "black box" in the decision-making process, lacking transparency. This lack of transparency becomes an obstacle when it's necessary to explain the basis of the answer or perform error analysis. Limited by the current structure of deep learning models, the output of large models may not match reality. While large models perform well in language understanding, they may not provide sufficiently deep or accurate answers when dealing with complex, multi-layered meanings or questions requiring in-depth domain knowledge.

[0011] Therefore, how to provide a hot topic question-answering implementation method that can update hot topic data in real time, has high interpretability and transparency, can overcome the large model illusion problem, and deeply understands the context is an urgent technical problem to be solved. Summary of the Invention

[0012] In view of this, embodiments of the present invention provide a hot topic question and answer assistant system and a hot topic question and answer implementation method based on the hot topic question and answer assistant system, so as to eliminate or improve one or more defects existing in the prior art.

[0013] One aspect of the present invention provides a method for implementing hot topic question answering based on a hot topic question answering assistant system. The method includes the following steps: parsing the original questions from users to extract key information points; constructing a thought chain based on all extracted key information points; integrating the constructed thought chain to obtain a question set; extracting feature vectors from the question set; using vector similarity analysis to match the question set with the most similar one from all preset hot topic knowledge bases as the target hot topic knowledge base for question retrieval; wherein the hot topic knowledge base is divided according to different event types; for each question in the question set, retrieving a sub-result from the target hot topic knowledge base using a multi-way retrieval strategy in a retrieval enhancement manner; integrating all sub-results obtained from all questions in the question set using the multi-way retrieval strategy to obtain a comprehensive question retrieval result; validating the comprehensive question retrieval result; and outputting the validated comprehensive question retrieval result.

[0014] In some embodiments of the present invention, the method further includes: pre-creating hotspot databases according to different event types, creating indexes for the hotspot databases; collecting text data about hotspot events at a preset frequency, performing preprocessing including text cleaning on the collected text data to obtain simplified event data, updating the simplified event data to the corresponding event type hotspot database at a preset frequency, and updating the indexes in the hotspot database.

[0015] In some embodiments of the present invention, the index of the hotspot database is implemented based on inverted index and TF-IDF information retrieval rules; the text cleaning includes: removing noise from the text data using regular expressions and natural language processing techniques; unifying the vocabulary contained in the text data to its basic form through stemming and lemmatization techniques; and identifying and merging synonyms in the text data using semantic similarity algorithms.

[0016] In some embodiments of the present invention, a thought chain is constructed based on all the extracted key information points, and the constructed thought chain is integrated to obtain a question set, including: generating related questions based on the extracted key information points; expanding the original questions and related questions in context based on the key information points to obtain a question set.

[0017] In some embodiments of the present invention, the method of using vector similarity analysis to match the question set with the one with the highest similarity among all preset hot topic knowledge bases as the target hot topic knowledge base for question retrieval includes: performing vector similarity calculation between the extracted feature vector of the question set and the feature vector of each pre-calculated hot topic knowledge base, taking the one with the highest similarity calculation result as the hot topic knowledge base with the highest similarity, and selecting the hot topic knowledge base of the corresponding event type as the target hot topic knowledge base for question retrieval.

[0018] In some embodiments of the present invention, for each question included in the question set, a retrieval sub-result is obtained by retrieving it from the target hot topic knowledge base using a multi-way retrieval strategy in a retrieval enhancement manner. Integrating all retrieval sub-results obtained from all questions in the question set using the multi-way retrieval strategy includes: inputting each question in the question set into the target hot topic knowledge base, returning a set of retrieval sub-results respectively using a multi-way retrieval strategy in a retrieval enhancement manner; unifying the text form of each retrieval sub-result using motion extraction and lexical reconstruction methods; using a semantic similarity algorithm to identify and merge synonyms in each retrieval sub-result, so as to integrate all retrieval sub-results into a single comprehensive question retrieval result.

[0019] In some embodiments of the present invention, the dimensions of the result verification include logical errors, common sense errors, and factual errors. When the result verification fails, the original question containing the reasons for the failure of the result verification is added and re-entered into the hot topic Q&A assistant system.

[0020] Corresponding to the above methods, the present invention also provides a hot topic question and answer assistant system, including a processor, a memory, and a computer program / instructions stored in the memory. The processor is used to execute the computer program / instructions. When the computer program / instructions are executed, the hot topic question and answer assistant system implements the steps of any of the methods described in the above embodiments.

[0021] In accordance with the above methods, the present invention also provides a computer-readable storage medium having a computer program / instructions stored thereon, which, when executed by a processor, implements the steps of the method as described in any of the above embodiments.

[0022] Corresponding to the above methods, the present invention also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the method as described in any of the above embodiments.

[0023] The hot topic question-and-answer assistant system and its implementation method proposed in this invention introduce a thought chain technique to decompose questions. It uses similarity analysis to match the decomposed questions with appropriate target hot topic knowledge bases and verifies the results to ensure the quality of the hot topic answers to users' original questions, thus enabling the understanding and support of complex business processes. Furthermore, this method stores event-related data for different event types separately, which helps reduce the amount of data retrieved from the database in the early stages. The multi-path retrieval strategy effectively improves the recall rate of the question-and-answer method, enhancing the credibility and effectiveness of the retrieval results.

[0024] Additional advantages, objects, and features of the invention will be set forth in part in the description which follows, and will also become apparent in part to those skilled in the art upon studying the description, or may be learned by practice of the invention. The objects and other advantages of the invention can be realized and obtained by means of the structures specifically pointed out in the description and drawings.

[0025] Those skilled in the art will understand that the objectives and advantages achievable with the present invention are not limited to those specifically described above, and that the above and other objectives achievable with the present invention will become clearer from the following detailed description. Attached Figure Description

[0026] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, are not intended to limit the scope of the invention. In the drawings:

[0027] Figure 1 This is a flowchart of a hot topic question-and-answer implementation method in one embodiment of the present invention.

[0028] Figure 2 This is a flowchart illustrating a specific implementation method for hot topic Q&A in another embodiment of the present invention.

[0029] Figure 3 This is a schematic diagram of the structure of the computer device included in the hot topic question and answer assistant system in one embodiment of the present invention. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.

[0031] It should also be noted that, in order to avoid obscuring the invention with unnecessary details, only the structures and / or processing steps closely related to the solution according to the invention are shown in the accompanying drawings, while other details that are not closely related to the invention are omitted.

[0032] It should be emphasized that the term "including / comprises" as used herein refers to the presence of a feature, element, step, or component, but does not exclude the presence or addition of one or more other features, elements, steps, or components.

[0033] It should also be noted that, unless otherwise specified, the term "connection" in this article can refer not only to a direct connection, but also to an indirect connection involving an intermediary.

[0034] In the following description, embodiments of the invention will be illustrated with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar parts, or the same or similar steps.

[0035] To overcome the problems existing in the prior art, this invention proposes a hot topic question-and-answer assistant system and a hot topic question-and-answer implementation method based on the hot topic question-and-answer assistant system. From a fundamental principle perspective, the hot topic question-and-answer assistant system and implementation method comprise two main aspects: In terms of data, data of different event types is uploaded to separate databases and updated daily with regular index updates; in terms of assistant execution, the question-and-answer operation is broken down into several sub-steps: related question generation, candidate knowledge base identification, related knowledge retrieval, and answer authenticity verification. All of these sub-steps are implemented using a thought chain approach to ensure the completeness and authenticity of the output results.

[0036] Figure 1 This is a flowchart of a hot topic question-and-answer implementation method according to an embodiment of the present invention. The hot topic question-and-answer implementation method based on a hot topic question-and-answer assistant system includes the following steps:

[0037] Step S110: Analyze the original questions from users to extract the key information points of the original questions, construct a thought chain based on all the extracted key information points, and integrate the constructed thought chains to obtain a question set.

[0038] Among them, the Chain of Thought (CoT) method breaks down complex problems into a series of smaller, more manageable subproblems. By generating the solution steps for these subproblems, the model can progressively construct the final answer. This approach not only improves the accuracy of the answer but also enhances the interpretability of the reasoning process.

[0039] Step S120: Extract the feature vector of the question set, and use vector similarity analysis to match the question set with the one with the highest similarity among all preset hot topic knowledge bases, which is used as the target hot topic knowledge base for question retrieval; wherein, the hot topic knowledge base is divided according to different event types.

[0040] Step S130: For each question in the question set, a sub-result is obtained by searching the target hot topic knowledge base using a multi-way search strategy in a search enhancement manner. All sub-results obtained by the multi-way search strategy for all questions in the question set are integrated to obtain a comprehensive question search result.

[0041] Step S140: Perform result verification on the comprehensive query retrieval results and output the verified comprehensive query retrieval results.

[0042] The hot topic question-answering implementation method proposed in this invention, based on a hot topic question-answering assistant system, can introduce thinking chain technology to decompose questions. It uses similarity analysis to match the decomposed questions with appropriate target hot topic knowledge bases and can verify the results to ensure the quality of hot topic answers to users' original questions, thus enabling understanding and support for complex business processes. Furthermore, this method stores event-related data for different event types separately, which helps reduce the amount of data retrieved from the database in the early stages. The multi-path retrieval strategy can effectively improve the recall rate of the question-answering method and enhance the credibility and effectiveness of the retrieval results.

[0043] In some embodiments of the present invention, the method further includes: (1) creating hotspot databases in advance according to different event types and creating indexes for the hotspot databases; (2) collecting text data about hotspot events at a preset frequency, performing preprocessing including text cleaning on the collected text data to obtain simplified event data, updating the simplified event data to the hotspot database of the corresponding event type at a preset frequency, and updating the indexes in the hotspot database.

[0044] In practice, a database management system (DBMS) can be selected to create a hotspot database. Based on the characteristics of the hotspot data, a reasonable database structure can be designed. This includes the design of data tables, fields, and indexes to optimize query performance. Event types can include breaking news, sporting events, and technology news.

[0045] By employing this invention, the amount of data retrieved from the database can be reduced in the early stages by creating separate hotspot databases according to different event types, significantly improving retrieval speed. Text cleaning and other operations ensure data quality and consistency, and the database can be automatically updated in a timely manner to guarantee its reliability. Through the establishment of an efficient real-time information capture and update mechanism, the system can quickly acquire and understand the latest hotspot information.

[0046] In some embodiments of the present invention, the index of the hotspot database is implemented based on the inverted index and TF-IDF information retrieval rules.

[0047] In some embodiments of the present invention, text cleaning includes: removing noise from text data using regular expressions and natural language processing techniques; unifying the vocabulary contained in the text data to its basic form through stemming and lemmatization techniques; and identifying and merging synonyms in the text data using semantic similarity algorithms.

[0048] By employing this embodiment of the invention, the quality of event data can be improved through noise removal, and the conciseness of event data can be ensured through vocabulary unification and synonym merging, thereby further improving the quality of event data.

[0049] In some embodiments of the present invention, a thought chain is constructed based on all the extracted key information points, and the constructed thought chain is integrated to obtain a question set, including: (1) generating related questions based on the extracted key information points; (2) expanding the original questions and related questions in context based on the key information points to obtain a question set.

[0050] In the specific implementation process, the thought chain can be automatically constructed through NLP models, the question can be enhanced by text augmentation using NLP models, and key information points can be extracted and thought chains (in the form of knowledge graphs) can be constructed by automatically constructing event knowledge graphs.

[0051] This invention utilizes mind chain technology to break down a user's original question. This facilitates a better understanding of the problem and allows for better questioning of a larger model and matching with relevant target hot topic knowledge bases, resulting in more comprehensive results. Furthermore, functions such as answer verification also rely on mind chain technology. Moreover, using mind chain technology helps improve the explainability and transparency of this hot topic question-and-answer implementation scheme.

[0052] In some embodiments of the present invention, the method of using vector similarity analysis to match the question set with the one with the highest similarity among all preset hot topic knowledge bases as the target hot topic knowledge base for question retrieval includes: performing vector similarity calculation between the extracted feature vector of the question set and the feature vector of each pre-calculated hot topic knowledge base, taking the one with the highest similarity calculation result as the hot topic knowledge base with the highest similarity, and selecting the hot topic knowledge base of the corresponding event type as the target hot topic knowledge base for question retrieval.

[0053] By employing this embodiment of the invention, a corresponding target knowledge base can be selected first based on semantic similarity. Knowledge retrieval within this target knowledge base significantly reduces the computational complexity in the initial stages and improves retrieval efficiency. In existing technologies, the increasing scale of data makes the recall of search results more difficult. This method stores various events in different knowledge bases, first identifying the user's intent based on the question before retrieval, and then performing content retrieval within the matched knowledge base.

[0054] In some embodiments of the present invention, for each question contained in the question set, a retrieval sub-result is obtained by using a multi-way retrieval strategy in the target hot topic knowledge base according to the retrieval enhancement method. The retrieval sub-results obtained by using the multi-way retrieval strategy for all questions contained in the question set are integrated, including: (1) inputting each question contained in the question set into the target hot topic knowledge base, and returning a set of retrieval sub-results respectively by using a multi-way retrieval strategy in the retrieval enhancement method; (2) unifying the text form of each retrieval sub-result by using motion extraction and word form restoration, and using a semantic similarity algorithm to identify synonyms in each retrieval sub-result and merge them, so as to integrate all retrieval sub-results into a comprehensive question retrieval result.

[0055] This invention improves retrieval quality. Firstly, in terms of data storage, data is maintained in different knowledge bases based on different event types. Secondly, in terms of retrieval, keyword and semantic similarity algorithms are used for multi-path result recall to ensure data integrity and comprehensiveness. These solutions guarantee high system availability and are the core technology of the system. Multiple rounds of retrieval verify the authenticity of the answers: providing retrieval results ensures that the generated answers are authentic and reliable, minimizing the illusion problem caused by large models.

[0056] In some embodiments of the present invention, the dimensions of the result verification include logical errors, common sense errors, and factual errors. When the result verification fails, the original question containing the reasons for the failure of the result verification is added and re-entered into the hot topic Q&A assistant system.

[0057] By employing this embodiment of the invention, it is possible to avoid incorrect expression or lack of common sense in the output question and answer results. At the same time, by comparing with hot data contained in the hot data database or the original text data associated with hot data, it is possible to determine whether there are factual errors, thereby ensuring the credibility of the output results.

[0058] Figure 2 This is a flowchart illustrating a specific implementation method for hot topic question answering in another embodiment of the present invention. This embodiment can be specifically broken down into the following 5 steps.

[0059] Step 1: Text cleaning and event storage.

[0060] To ensure the accuracy and efficiency of search results, the development of the hot topic Q&A assistant tool adopted a strategy of storing data in separate databases for different event types. The core of this strategy is that by classifying and storing data according to event type (such as breaking news, sporting events, and technology news), search speed can be significantly improved while ensuring the relevance and timeliness of the returned results. Before the data was stored, we implemented a series of meticulous text cleaning operations to ensure data quality and consistency.

[0061] Text cleaning is a crucial step in data preprocessing, encompassing the removal of irrelevant information, standardization of text formatting, correction of spelling errors, elimination of duplicate data, and handling of missing values. Specifically, we first use regular expressions and natural language processing techniques to remove noise from the text, such as HTML tags, special characters, and meaningless punctuation. Next, we use stemming and lemmatization techniques to unify words to their basic forms, reducing the impact of word variations. Furthermore, we utilize semantic similarity algorithms to merge synonyms, further improving data consistency. After processing the text content, we also check and handle missing values ​​in the data to ensure the integrity of each record.

[0062] After text cleaning, the data is categorized and stored in the corresponding event type database. This process involves not only data classification but also index building to accelerate subsequent retrieval operations. We used inverted indexes and the TF-IDF algorithm to generate keyword indexes for each record, enabling the system to quickly locate the relevant event database and retrieve the most matching record when a user submits a query. Through this series of text cleaning and event storage operations, our hot topic Q&A assistant tool can provide more accurate and faster retrieval services, meeting users' immediate needs for information on trending events.

[0063] Step 2: Enhance user questions using mind chain technology.

[0064] After receiving user questions, using the thought chain approach to abstract the questions and extract more related questions is a strategy to enhance understanding and generate more comprehensive and in-depth answers.

[0065] The system needs to analyze the user's question, identifying key information points such as topic, time, location, and people involved. Based on these key information points, the system constructs a thought chain, including: 1. Related question generation: generating a series of related questions from different perspectives to explore multiple aspects of the problem. 2. Context expansion: considering the context of the problem, such as historical background and social environment, to provide richer information. Finally, the information obtained from the thought chain is integrated to construct a comprehensive framework for understanding the problem.

[0066] Answer generation: Based on the integrated information, generate one or more answers that not only directly answer the user's question, but may also include additional, useful information related to the question.

[0067] Step 3: Knowledge base matching based on vector similarity.

[0068] To avoid increasing the difficulty of knowledge retrieval due to the growing data scale, different types of events need to be stored in separate databases according to event type. An enhanced set of user questions is used for intent recognition to determine which knowledge base the user expects the answer to be in. A vector similarity-based scheme is used to compare the maintained knowledge base with the user questions, selecting the most similar knowledge base for retrieval. Vector similarity is an indicator that measures the degree of similarity between two vectors in a vector space. In mathematics and computer science, vectors are typically composed of an ordered set of values ​​and can represent various types of data, such as text, images, and sound. Vector similarity helps us understand the degree of similarity between different data points. Clustering algorithms based on vector similarity can group similar data points together.

[0069] Step 4: RAG-based multi-path retrieval recall.

[0070] To address the complexity and diversity challenges faced by information retrieval and recommendation systems, a multi-strategy approach is adopted for query retrieval, employing multiple strategies for combined recall. By combining various strategies, more comprehensive relevant information can be retrieved, improving recall and ensuring users obtain as many potentially relevant results as possible. A single retrieval strategy may perform poorly in certain situations, such as when the query semantics are ambiguous or the data distribution is uneven. Multiple strategies can complement each other, enhancing the system's robustness and ensuring high-quality retrieval results across different scenarios. Recall, a concept from machine learning, refers to the proportion of samples correctly identified as positive by the model out of all actual positive samples. It reflects the model's coverage or ability to capture positive samples. Retrieval Augmented Generation (RAG) refers to a model first retrieving the most relevant information to the input query from a large-scale document collection or knowledge base, and then using this retrieved information to enhance the model's output, thereby improving the accuracy and relevance of the generated content.

[0071] Step 5: Validate the generated results.

[0072] To avoid the "illusion" problem that occurs when large models (such as deep learning models, especially pre-trained language models) generate results (i.e., the information generated by the model does not match the facts, or contains logical errors, common sense errors, etc.), result verification is performed after the results are generated.

[0073] Corresponding to the above method, the present invention also provides a hot topic question and answer assistant system, which includes a computer device, the computer device including a processor and a memory, the memory storing computer instructions, the processor being used to execute the computer instructions stored in the memory, and when the computer instructions are executed by the processor, the hot topic question and answer assistant system implements the steps of the method described above.

[0074] Figure 3 This is a schematic diagram of the structure of the computer device included in the hot topic question and answer assistant system in one embodiment of the present invention.

[0075] See Figure 3 The computer device 00 includes: a processor 01, a memory 02, and a computer program stored on the memory 02 and executable on the processor 01. When the processor 01 executes the computer program, it implements the human factors data server access control method provided in the above method embodiments.

[0076] The processor 01 is connected to the memory 02, such as via a bus 03. The processor 01 can be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. The processor 01 can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc. The bus 03 may include a pathway for transmitting information between the aforementioned components. The bus 03 can be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 130 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 The text uses only a single thick line to represent a bus, but this does not imply that there is only one bus or one type of bus. Memory 02 stores a computer program corresponding to the human factors data server access control method described in the above embodiments of this application. This computer program is executed under the control of processor 01. Processor 01 executes the computer program stored in memory 02 to implement the content shown in the aforementioned method embodiments.

[0077] Corresponding to the methods described above, the present invention also provides a computer-readable storage medium having a computer program / instructions stored thereon, which, when executed by a processor, implements the steps of the method as described in any of the above embodiments. The computer-readable storage medium may be a tangible storage medium, such as random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, register, floppy disk, hard disk, removable storage disk, CD-ROM, or any other form of storage medium known in the art.

[0078] Corresponding to the above methods, the present invention also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the method as described in any of the above embodiments.

[0079] Those skilled in the art will understand that the exemplary components, systems, and methods described in conjunction with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Whether implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention. When implemented in hardware, it can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this invention are programs or code segments used to perform the desired tasks. The programs or code segments can be stored in a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried in a carrier wave.

[0080] It should be clarified that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of the present invention.

[0081] In this invention, features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, and / or combined with or in place of features of other embodiments.

[0082] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, various modifications and variations of the embodiments of the present invention are possible. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for implementing hot topic question and answer based on a hot topic question and answer assistant system, characterized in that, include: The original questions from users are analyzed to extract the key information points. Based on all the extracted key information points, a thought chain is constructed, and the constructed thought chain is integrated to obtain a set of questions. The feature vectors of the question set are extracted, and the question set is matched with the one with the highest similarity in the entire preset hot topic knowledge base using vector similarity analysis. This hot topic knowledge base is used as the target hot topic knowledge base for question retrieval. The hot topic knowledge base is divided according to different event types. For each question in the question set, a sub-result is obtained by searching the target hot topic knowledge base using a multi-way search strategy in a search enhancement manner. All sub-results obtained by searching all questions in the question set using the multi-way search strategy are integrated to obtain a comprehensive question search result. The query retrieval results are validated, and the validated query retrieval results are output.

2. The method according to claim 1, characterized in that, The method further includes: Create separate hotspot databases for different event types in advance, and create indexes for the hotspot databases; Text data about hot events is collected at a preset frequency. The collected text data is preprocessed, including text cleaning, to obtain concise event data. The concise event data is then updated to the hot event database of the corresponding event type at a preset frequency, and the index in the hot event database is also updated.

3. The method according to claim 2, characterized in that, The index of the hotspot database is implemented based on inverted indexes and TF-IDF information retrieval rules; The text cleaning process includes: removing noise from the text data using regular expressions and natural language processing techniques; unifying the vocabulary in the text data to its basic form through stemming and lemmatization techniques; and identifying and merging synonyms in the text data using semantic similarity algorithms.

4. The method according to claim 1, characterized in that, Based on all the extracted key information points, a thought process chain is constructed. Integrating this chain yields a set of questions, including: Generate relevant questions based on the extracted key information points; Based on key information points, the original question and related questions are expanded in context to obtain a set of questions.

5. The method according to claim 1, characterized in that, Vector similarity analysis is used to match the question set with the most similar question from all preset hot topic knowledge bases. This hot topic knowledge base is then used as the target for question retrieval. The feature vectors of the extracted question set are compared with the feature vectors of each pre-calculated hot topic knowledge base. The hot topic knowledge base with the highest similarity is selected as the hot topic knowledge base with the highest similarity. The hot topic knowledge base with the corresponding event type is selected as the target hot topic knowledge base for question retrieval.

6. The method according to claim 1, characterized in that, For each question in the question set, a sub-result is retrieved from the target hot topic knowledge base using a multi-way retrieval strategy in a retrieval enhancement manner. All sub-results obtained from all questions in the question set using the multi-way retrieval strategy are then integrated, including: Each question in the question set is input into the target hot topic knowledge base, and a set of search sub-results are returned respectively using a multi-way search strategy in a search enhancement manner; The text format of each search sub-result is unified by using motion extraction and word form restoration. A semantic similarity algorithm is used to identify synonyms in each search sub-result and merge them to integrate all search sub-results into a single comprehensive query search result.

7. The method according to claim 1, characterized in that, The dimensions of result verification include logical errors, common sense errors, and factual errors. When the result verification fails, the original question containing the reason for the failure is added and re-entered into the hot topic Q&A assistant system.

8. A hot topic question-and-answer assistant system, comprising a processor, a memory, and a computer program / instructions stored in the memory, characterized in that, The processor is used to execute the computer program / instructions, and when the computer program / instructions are executed, the hot topic question and answer assistant system implements the steps of the method as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method as described in any one of claims 1 to 7.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1 to 7.

Images