Data query method and device, equipment, storage medium and computer program product

Abstract

The invention provides a data query method, device and equipment, a storage medium and a computer program product, and relates to the technical field of computers.The method comprises the steps that input data of a user is received, and data arrangement is conducted on the input data; inputting the sorted data into a professional knowledge model to obtain explanatory data output by the professional knowledge model, and inputting the explanatory data into a language understanding model to obtain an understanding result of the explanatory data output by the language understanding model; and under the condition that the understanding result is understanding interpretation data, outputting the interpretation data as a query result. Through intelligent knowledge arrangement and common knowledge expression, a query result easy to understand by a user is output, so that the user is helped to deeply understand knowledge in a specific field in a short time.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a data query method, device, equipment, storage medium and computer program product. Background Art

[0002] With the rapid development of information technology, the amount of information has exploded, and it has become increasingly difficult for people to master new knowledge in a short period of time. Traditional learning platforms rely on user self-learning and lack effective auxiliary tools to help understand complex concepts or large amounts of information. This makes learners feel overwhelmed when faced with complex knowledge and difficult to gain a deep understanding. In addition, although the existing intelligent question-answering system can provide information retrieval, it is still insufficient in the deep understanding and popular expression of knowledge, which limits the user's learning experience and effect. Therefore, how to provide users with easy-to-understand query results in a short period of time has become an urgent problem to be solved. Summary of the invention

[0003] The present invention provides a data query method, apparatus, device, storage medium and computer program product to solve the defects in the prior art of how to provide users with easy-to-understand query results in a short time, and output query results that are easy for users to understand through intelligent knowledge organization and popular knowledge expression, thereby helping users to deeply understand the knowledge in a specific field in a short time.

[0004] The present invention provides a data query method, comprising the following steps: Receiving input data from a user, and collating the input data; Inputting the sorted data into a professional knowledge model to obtain interpretation data output by the professional knowledge model, and inputting the interpretation data into a language understanding model to obtain an understanding result of the interpretation data output by the language understanding model; When the understanding result is understanding of the explanation data, the explanation data is output as a query result.

[0005] According to a data query method provided by the present invention, the step of sorting the input data includes: determining the type of the input data; If the type of the input data is a concept, then extracting the associated data of the concept from the knowledge base, and logically sorting the key data in the associated data; If the type of the input data is a book, the book is decomposed according to chapters to determine chapter data of each chapter, and the chapter data are logically sorted according to the chapters of the book; If the type of the input data is a paper, the paper is summarized in sections to determine summary data for each paragraph, and the summary data is logically sorted according to the paragraph information of the paper.

[0006] According to a data query method provided by the present invention, extracting the associated data of the concept from the knowledge base includes: Convert the concept into a vectorized representation to obtain a query vector; Performing preliminary indexing on the query vector to obtain preliminary index data; The preliminary index data is subjected to secondary indexing to obtain the associated data.

[0007] According to a data query method provided by the present invention, the knowledge base is constructed based on the following method: Extract preprint data from at least one data source; According to the type of the preprint data, vectorizing the preprint data; The knowledge base is constructed based on the vectorized preprint data.

[0008] According to a data query method provided by the present invention, after inputting the sorted data into a professional knowledge model to obtain interpretation data output by the professional knowledge model, and inputting the interpretation data into a language understanding model to obtain an understanding result of the interpretation data output by the language understanding model, the method further comprises: When the understanding result is that the interpretation data is not understood, the understanding result is input into the professional knowledge model, so that the professional knowledge model performs iterative interpretation based on the understanding result until the interpretation data understood by the language understanding module is output.

[0009] According to a data query method provided by the present invention, when the understanding result is understanding the explanation data, after outputting the explanation data as the query result, the method further comprises: Obtaining the scoring results of the explained data in multiple evaluation dimensions, as well as the user's correction suggestions for the explained data; The knowledge base and the knowledge interpretation algorithm are optimized according to the scoring results and the correction suggestions.

[0010] The present invention also provides a data query device, comprising the following modules: A data sorting module, used for receiving input data from a user and sorting the input data; An interpretation module, used to input the sorted data into a professional knowledge model to obtain interpretation data output by the professional knowledge model, and input the interpretation data into a language understanding model to obtain an understanding result of the interpretation data output by the language understanding model; The output module is used to output the explanation data as a query result when the understanding result is understanding the explanation data.

[0011] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements any of the above-mentioned data query methods when executing the computer program.

[0012] The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the data query method described above is implemented.

[0013] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above-mentioned data query methods.

[0014] The data query method, device, equipment, storage medium and computer program product provided by the present invention receive input data from a user, organize the input data, input the organized data into a professional knowledge model to obtain interpretation data output by the professional knowledge model, and input the interpretation data into a language understanding model to obtain an understanding result of the interpretation data output by the language understanding model; when the understanding result is understanding of the interpretation data, the interpretation data is output as a query result. The present invention outputs query results that are easy for users to understand through intelligent knowledge organization and popular knowledge expression, thereby helping users to deeply understand knowledge in a specific field in a short time. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 It is a flow chart of the data query method provided by the present invention.

[0017] Figure 2 It is a flow chart of the intelligent interactive rapid learning method based on the knowledge base provided by the present invention.

[0018] Figure 3It is a structural schematic diagram of the data query device provided by the present invention.

[0019] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] Combine the following Figure 1-Figure 4 The data query method, apparatus, device, storage medium and computer program product of the present invention are described.

[0022] Figure 1 It is a flow chart of the data query method provided by the present invention, such as Figure 1 As shown, the method includes the following: Step 101, receiving input data from a user, and sorting the input data.

[0023] It should be noted that the executor of the present invention is a data query system, specifically an intelligent interactive rapid learning system based on a knowledge base, which can help users deeply understand the knowledge in a specific field in a short time through intelligent knowledge organization and popular knowledge expression.

[0024] When users want to obtain specific information or answer questions, they can input data into the system, including text data, such as query questions or concepts; file data, such as pictures, charts, and uploaded documents (such as PDF, Word), etc. Then, the system organizes the user's input data.

[0025] In one embodiment, data sorting of user input data includes: determining the type of input data; if the type of input data is a concept, extracting associated data of the concept from the knowledge base, and logically sorting the key data in the associated data; if the type of input data is a book, decomposing the book according to chapters to determine chapter data of each chapter, and logically sorting the chapter data according to the chapters of the book; if the type of input data is a paper, summarizing the paper in sections to determine summary data of each paragraph, and logically sorting the summary data according to the paragraph information of the paper.

[0026] For example, when a user proposes a concept, book, or paper, the system adopts a corresponding organization strategy to ensure that the final content structure is helpful for the user's understanding. The organization strategies are as follows: (1) Concept organization: If the user inputs a specific concept, the system will directly extract the information related to the concept from the knowledge base, and then logically sort and refine the extracted information to form an easy-to-understand explanation structure. This process can include a progressive explanation of the concept definition, background information, related theories, and practical applications.

[0027] (2) Book organization: If the user inputs a book, the system will first automatically summarize and organize the entire book. The system will break down the book into chapters, and each chapter will be summarized independently to extract the key ideas, main arguments, and important data. Then, the system will arrange these chapter summaries in a logical order to form an overview of the entire book. At the same time, the system will decide whether certain parts need to be further refined or simplified based on the complexity of the book content and the user's understanding needs to ensure the clarity and comprehensibility of the final explanation.

[0028] (3) Paper Arrangement: If the user inputs a paper, the system will summarize the paper in sections, including independent summaries of the abstract, introduction, methods, results, and conclusions. In the process of summarizing, the system extracts the core findings, research methods, and main conclusions of the paper. Then, the system will arrange these summaries according to the logical structure of the paper and further explain some complex parts when necessary.

[0029] In the process of organizing and summarizing concepts, books, or papers, the system dynamically extracts auxiliary knowledge related to the current explanation from the knowledge base. This on-the-fly retrieval method ensures that the explanation of each concept or chapter can be fully supported and expanded. For example, when explaining a complex concept, the system may retrieve related diagrams, analogies, or supplementary information in other fields to enhance the clarity and comprehensiveness of the explanation.

[0030] The embodiment of the present invention dynamically extracts auxiliary knowledge according to user input and generates a content structure that is most helpful for understanding, thereby ensuring that the user obtains the most relevant information.

[0031] Step 102, inputting the sorted data into the professional knowledge model to obtain the interpretation data output by the professional knowledge model, and inputting the interpretation data into the language understanding model to obtain the understanding result of the interpretation data output by the language understanding model.

[0032] Step 103: when the understanding result is understanding of the explanation data, output the explanation data as a query result.

[0033] It should be noted that the professional knowledge model not only has a deep professional knowledge background, but also can explain complex concepts in a concise and clear way. The language understanding model is used to understand the professional knowledge expressed by the professional knowledge model in plain and easy-to-understand words and provide feedback to the professional knowledge model.

[0034] During system operation, the professional knowledge model first attempts to explain the concepts proposed by the user in plain language, and then passes this explanation to the language understanding model. If the language understanding model can understand the explanation data output by the professional knowledge model, the explanation data is directly output as the query result. If the language understanding model does not understand the explanation data output by the professional knowledge model, the understanding result is input into the professional knowledge model so that the professional knowledge model iteratively interprets based on the understanding result until the explanation data understood by the language understanding module is output. For example, if the language understanding model fails to fully understand or has doubts, the professional knowledge model will further explain in depth, or use various materials in the knowledge base (such as text, pictures, videos, etc.) to assist in the explanation until the language understanding model fully understands the concept.

[0035] It is understandable that the understanding results output by the language understanding model may include: The above is the explanation of {concept} by the professional knowledge model {output content of the professional knowledge model}. If you can understand this explanation, please output {I understand}; if there is part that you do not understand, please reply {I generally understand, but the specific [concept] part is not clear, please further explain or provide more relevant information.}. Based on this, the feedback from the language understanding model will include unclear concepts and parts that need further clarification. For example: "I generally understand, but 'X concept' and 'Y concept' are not very clear. Please re-explain these two concepts in detail or add more details to help understand."

[0036] To ensure that the language understanding model can grasp complex knowledge, the system uses some prompt words during the interaction process, such as: "Please explain [concept] in a way that a beginner with no basic knowledge can understand." Or "In my explanation of [concept], identify potential loopholes or unclear parts: [explain]." In addition, the system can also ask the professional knowledge model to "add two analogies to help better understand [concept]." Or "Draw a diagram to help understand [concept]." Through this iterative explanation and feedback mechanism, the system can eventually output an explanation that ordinary users can easily understand. This repeated interactive process not only ensures the accuracy of the explanation, but also greatly improves the comprehensibility of knowledge.

[0037] The data query method provided by the embodiment of the present invention receives input data from a user and organizes the input data; inputs the organized data into a professional knowledge model to obtain interpretation data output by the professional knowledge model, and inputs the interpretation data into a language understanding model to obtain an understanding result of the interpretation data output by the language understanding model; when the understanding result is understanding of the interpretation data, the interpretation data is output as a query result. The present invention outputs query results that are easy for users to understand through intelligent knowledge organization and popular knowledge expression, thereby helping users to deeply understand the knowledge in a specific field in a short time.

[0038] Based on the above embodiment, extracting the associated data of the concept from the knowledge base includes: Step 210, converting the concept into a vectorized representation to obtain a query vector; Step 211, performing preliminary indexing on the query vector to obtain preliminary index data; Step 212: perform secondary indexing on the preliminary index data to obtain the associated data.

[0039] The user inputs concepts through the system interface, and the system converts the concepts input by the user into vector representations. For example, the same natural language processing technology as the knowledge base is used to ensure that the query and the stored content are compatible in the vector space. Then, the retrieval capability of the Milvus vector database is used for preliminary indexing. Based on the vector queried by the user, similar vectors are quickly found. Milvus will return several candidates with the highest similarity to the query vector. After obtaining the candidates, the RAPTOR indexing method is used for more refined retrieval, i.e., secondary indexing. The system further processes the candidates through a tree structure to extract more specific and relevant information. The search results are further integrated and sorted based on similarity and relevance. For example, the system will combine additional metadata (such as article publication time, number of citations, etc.) to optimize the result display.

[0040] Optionally, other data can be extracted based on the knowledge base. Suppose the user is interested in "Application of deep learning in image recognition" and wants to obtain relevant research papers. The user can enter the query "deep learning image recognition" in the system interface. The system converts the user's query into a vector representation, for example, by converting the query text into a vector through a pre-trained natural language processing model, and then using the Milvus vector database, the system searches for vectors similar to the user's query vector. After obtaining a series of candidate results, the system further analyzes these candidates using the RAPTOR index to integrate information. RAPTOR extracts specific research content and application cases related to "deep learning" and "image recognition" from the tree structure. The system sorts these candidate papers according to relevance, and may take into account factors such as the number of citations and publication time. In the end, the user may see the following papers: "Application of deep learning in medical image analysis", "Recent advances in convolutional neural networks in object recognition" and "Technical review of image classification using deep learning". The system displays the selected paper abstracts, author information, links and other information to the user for further reading.

[0041] With the help of Milvus' high-performance retrieval capability and RAPTOR indexing method, the system of the embodiment of the present invention can quickly extract the most relevant content from the knowledge base according to the user's query requirements.

[0042] Based on the above embodiment, the knowledge base is constructed in the following manner: Step 220, extracting preprint data from at least one data source; Step 221, performing vectorization processing on the preprint data according to the type of the preprint data; Step 222: construct the knowledge base based on the vectorized preprint data.

[0043] In order to achieve efficient knowledge management and retrieval, the embodiment of the present invention uses a vector database based on Milvus to build a knowledge base, and combines it with the RAPTOR (Recursive Abstractive Processing for Tree-Organized Retrieval) indexing method for optimization. Among them, Milvus is a high-performance open source vector database suitable for processing large-scale, irregular unstructured data, such as text, images, and videos. Through Milvus, the data in the knowledge base is converted into vector representation, thereby supporting efficient semantic retrieval and similarity search.

[0044] A knowledge base is built based on preprint data from at least one data source. For example, the knowledge base is mainly built based on the arXiv Dataset. arXiv is a website that collects preprint papers in many fields, covering cutting-edge research results in physics, mathematics, computer science and other fields. Each preprint contains rich metadata, such as text content, charts, author information, citations, classifications, etc.

[0045] In the process of building the knowledge base, the system first extracts all relevant preprint articles from the arXiv Dataset and converts them into vector representations suitable for storage in the Milvus vector database. The specific steps are as follows: Data cleaning and preprocessing: The system cleans and preprocesses the text, graphics, and metadata in the arXiv Dataset to ensure data consistency and integrity.

[0046] Vectorization: Natural Language Processing (NLP) technology is used to convert text information into vector form. At the same time, the system uses computer vision technology to convert chart information into image vectors. Other metadata such as author information, citations, and classifications are also converted into appropriate vector representations.

[0047] Vector database construction: All vectorized data is stored in the Milvus vector database. Milvus supports the storage and retrieval of large-scale vector data, allowing the system to quickly find the content most relevant to user needs when querying.

[0048] RAPTOR indexing method: In order to further improve retrieval efficiency, the system adopts the RAPTOR indexing method. RAPTOR is a new indexing technology that builds a hierarchical tree structure through recursive clustering and text block summarization. The process includes embedding, clustering and summarizing text blocks into a tree structure. Nodes at different levels represent different levels of abstraction. During the reasoning process, RAPTOR is able to retrieve information from all levels of the tree, enabling the model to effectively integrate broader subject context and fine-grained details. This method not only improves retrieval speed, but also enhances the accuracy of the system when processing complex queries.

[0049] Fast response mechanism: With the high-performance retrieval capability of Milvus and the RAPTOR indexing method, the system can quickly extract the most relevant content from the knowledge base according to the user's query requirements. This mechanism ensures that the system can still provide real-time response when facing large-scale data.

[0050] The knowledge base provided by the embodiment of the present invention not only covers a wide range of industry fields and subject content, but also has powerful retrieval capabilities and efficient knowledge management functions. The combination of the Milvus vector database and the RAPTOR indexing method ensures that the system can quickly locate the knowledge required by users in massive data, providing sufficient data support for subsequent intelligent sorting and translation modules. This efficient knowledge base construction method enables the system to perform excellent performance when processing complex queries, and can provide users with accurate and timely knowledge acquisition services.

[0051] Based on the above embodiment, when the understanding result is understanding the explanation data, after outputting the explanation data as the query result, the method further includes: Step 310, obtaining the scoring results of the explanation data in multiple evaluation dimensions, and the user's correction suggestions for the explanation data; Step 311, optimizing the knowledge base and knowledge interpretation algorithm according to the scoring results and the correction suggestions.

[0052] The system will display the generated final explanation to the user and provide a series of interactive options, allowing users to provide feedback on the accuracy, comprehensibility and overall satisfaction of the explanation. The user feedback module is designed to continuously optimize the system's knowledge base and knowledge interpretation algorithm through the user's real experience, thereby improving the overall performance of the system and user experience. Among them, user feedback mainly covers the following aspects: Overall multi-dimensional feedback scoring system: After receiving the final explanation, users can provide feedback on the system output in multiple dimensions, including the following aspects: understandability, users can evaluate the clarity and popularity of the explanation, and point out whether there are difficult-to-understand terms, complex sentence structures, or other factors that make it difficult to understand; level of detail, users can evaluate the level of detail of the explanation, and point out whether there are parts that need more detailed explanations, or whether there are overly detailed and lengthy parts; relevance, users can evaluate the relevance of the explanation to the original query, and point out whether there is irrelevant or insignificant content. For the above aspects, users can quickly score the three dimensions of the explanation through the scoring system (such as 1-5 stars).

[0053] User correction suggestions: The system allows users to directly make correction suggestions. For example, users can submit more accurate or clearer explanations, and the system will use these suggestions as reference data for future optimization.

[0054] Feedback-driven dynamic adjustment: The system will continuously monitor user feedback data in the background and use this data to dynamically adjust the content of the knowledge base. For example, if the explanation of a concept is repeatedly reported as difficult to understand, the system will analyze user feedback and adjust the explanation algorithm to add more analogies, charts or other auxiliary explanations.

[0055] Personalized learning path optimization: The system gradually adjusts its explanation style to users based on historical feedback records to better suit users' understanding ability and learning habits. For some users, the system will simplify the explanation; for users who need a deeper understanding, the system will provide more detailed explanations and additional reference materials.

[0056] Long-term storage and analysis of feedback data: All user feedback data will be stored by the system for a long time and used for regular system performance analysis. The system will cluster and recognize patterns in a large amount of feedback data through machine learning algorithms to extract common problems and optimization directions. These analysis results will be used to guide future knowledge base updates and algorithm improvements.

[0057] Through the user feedback module, the system can achieve the following goals: Continuous optimization: User feedback is an important basis for improving system performance. By continuously collecting and analyzing feedback, the system can continuously improve the quality of the knowledge base and the accuracy of the interpretation algorithm.

[0058] Personalized user experience: By recording and analyzing user feedback data, the system can gradually understand the needs and preferences of each user, thereby providing a more personalized learning experience.

[0059] Dynamic update of knowledge base: User feedback is not only used to optimize the interpretation algorithm, but also to guide the content update of the knowledge base. Based on user feedback, the system will prioritize those knowledge points that are closely related to user needs to ensure that the knowledge base is always up-to-date and relevant.

[0060] Improve user participation: Allowing users to participate in the system optimization process not only improves user participation, but also enhances users' trust and reliance on the system.

[0061] The embodiments of the present invention continuously optimize the system's knowledge base and knowledge interpretation algorithm through real user experience, thereby improving the overall performance of the system and user experience.

[0062] In order to further analyze and illustrate the data query method proposed in the present invention, refer to the following examples.

[0063] The embodiment of the present invention specifically proposes an intelligent interactive rapid learning system based on a knowledge base and a method thereof. The system constructs an efficient knowledge base, combines intelligent organization and dynamic knowledge extraction, and uses the iterative interaction of two large models to output query results that are easy for users to understand. At the same time, through the user feedback mechanism, fast, accurate and personalized knowledge acquisition and learning experience are achieved. The system includes four core modules: a knowledge base construction module, an intelligent organization and dynamic knowledge extraction module, a knowledge interpretation module and a user feedback module. The main functions of each module are as follows: The knowledge base construction module is used to collect data, perform preprocessing and vectorization, build the Milvus vector database, and optimize the retrieval efficiency through the RAPTOR indexing method.

[0064] Intelligent organization and dynamic knowledge extraction module, which performs logical organization and summary based on the concepts, books or papers input by the user, dynamically extracts auxiliary knowledge, and generates the content structure that is most conducive to understanding.

[0065] The knowledge interpretation module interacts with the professional knowledge model and the language understanding model through repeated iterations to output explanations that users can easily understand.

[0066] User feedback module, which displays the final explanation and collects user feedback, and uses feedback data to dynamically adjust the knowledge base content and explanation algorithm to improve system performance.

[0067] refer to Figure 2 ,The workflow of the system can be divided into the following key modules, each of which works together to ensure efficient and accurate knowledge retrieval, organization, interpretation, and feedback optimization: (1) Knowledge base construction module: Data collection and preprocessing: The system extracts relevant preprint articles from the arXiv Dataset and performs data cleaning and preprocessing.

[0068] Vectorization: Use NLP and computer vision techniques to convert text and chart data into vector representations.

[0069] Milvus vector database construction: stores vectorized data and supports large-scale efficient retrieval.

[0070] RAPTOR indexing method: Using RAPTOR indexing technology, a hierarchical tree structure is constructed through recursive clustering and text summarization to improve retrieval efficiency and accuracy.

[0071] (2) Intelligent organization and dynamic knowledge extraction module: Concept organization: Logically sort and refine the concepts input by the user to form an easy-to-understand explanation structure.

[0072] Book organization: Break down and summarize the book by chapters to form an overall book overview.

[0073] Paper organization: Summarize each part of the paper in sections and arrange them according to a logical structure.

[0074] Auxiliary knowledge extraction: During the interpretation process, relevant knowledge is dynamically extracted from the knowledge base to support and expand the interpretation.

[0075] (3) Knowledge interpretation module: Expertise model: tries to explain concepts in plain language and interacts with language understanding models.

[0076] Language understanding model: Understands and provides feedback on the explanation of the professional knowledge model to ensure that the final explanation is clear and understandable.

[0077] Iterative explanation and feedback mechanism: Through repeated interactions, the final output is an explanation that users can easily understand.

[0078] (4) User feedback module: Multi-dimensional feedback rating system: Users rate the accuracy, comprehensibility, detail, and relevance of explanations.

[0079] User correction suggestions: Users can submit correction suggestions, which the system will use as optimization reference.

[0080] Feedback-driven dynamic adjustment: Dynamically adjust the knowledge base content and interpretation algorithm based on feedback data.

[0081] Personalized learning path optimization: Adjust the explanation style based on user feedback to provide a personalized learning experience.

[0082] Long-term storage and analysis of feedback data: Use machine learning to analyze feedback data to guide knowledge base updates and algorithm improvements.

[0083] The embodiment of the present invention constructs an efficient knowledge base and combines intelligent organization with dynamic knowledge extraction, uses the iterative interaction of two large models to output query results that are easy for users to understand, and at the same time achieves fast, accurate and personalized knowledge acquisition and learning experience through a user feedback mechanism.

[0084] The data query device provided by the present invention is described below. The data query device described below and the data query method described above can be referenced to each other.

[0085] refer to Figure 3 The data query device provided by the present invention includes a data sorting module 301, an interpretation module 302 and an output module 303.

[0086] The data sorting module 301 is used to receive the input data of the user and sort the input data; The interpretation module 302 is used to input the sorted data into the professional knowledge model to obtain the interpretation data output by the professional knowledge model, and input the interpretation data into the language understanding model to obtain the understanding result of the interpretation data output by the language understanding model; The output module 303 is used to output the explanation data as a query result when the understanding result is understanding the explanation data.

[0087] The data query device provided by the embodiment of the present invention receives input data from a user and organizes the input data; inputs the organized data into a professional knowledge model to obtain interpretation data output by the professional knowledge model, and inputs the interpretation data into a language understanding model to obtain an understanding result of the interpretation data output by the language understanding model; when the understanding result is understanding of the interpretation data, the interpretation data is output as a query result. The present invention outputs query results that are easy for users to understand through intelligent knowledge organization and popular knowledge expression, thereby helping users to deeply understand the knowledge in a specific field in a short time.

[0088] In one embodiment, the data sorting module 301 is specifically used to: Determine the type of the input data; if the type of the input data is a concept, extract the associated data of the concept from the knowledge base, and logically sort the key data in the associated data; if the type of the input data is a book, decompose the book according to chapters to determine the chapter data of each chapter, and logically sort the chapter data according to the chapters of the book; if the type of the input data is a paper, summarize the paper in sections to determine the summary data of each paragraph, and logically sort the summary data according to the paragraph information of the paper.

[0089] In one embodiment, the data sorting module 301 is specifically used to: The concept is converted into a vectorized representation to obtain a query vector; the query vector is preliminarily indexed to obtain preliminary index data; the preliminary index data is re-indexed to obtain the associated data.

[0090] In one embodiment, the data sorting module 301 is further used to: Extracting preprint data from at least one data source; vectorizing the preprint data according to the type of the preprint data; and constructing the knowledge base based on the vectorized preprint data.

[0091] In one embodiment, the interpretation module 302 is further configured to: When the understanding result is that the interpretation data is not understood, the understanding result is input into the professional knowledge model, so that the professional knowledge model performs iterative interpretation based on the understanding result until the interpretation data understood by the language understanding module is output.

[0092] In one embodiment, the output module 303 is further configured to: Obtaining the scoring results of the explained data in multiple evaluation dimensions, as well as the user's correction suggestions for the explained data; The knowledge base and the knowledge interpretation algorithm are optimized according to the scoring results and the correction suggestions.

[0093] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4 As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430 and a communication bus 440, wherein the processor 410, the communication interface 420 and the memory 430 communicate with each other through the communication bus 440. The processor 410 may call the logic instructions in the memory 430 to execute the data query method, which includes: receiving input data from a user, sorting the input data; inputting the sorted data into a professional knowledge model to obtain interpretation data output by the professional knowledge model, and inputting the interpretation data into a language understanding model to obtain an understanding result of the interpretation data output by the language understanding model; and outputting the interpretation data as a query result when the understanding result is understanding the interpretation data.

[0094] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0095] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the data query method provided by the above-mentioned methods, which method includes: receiving user input data and organizing the input data; inputting the organized data into a professional knowledge model to obtain interpretation data output by the professional knowledge model, and inputting the interpretation data into a language understanding model to obtain an understanding result of the interpretation data output by the language understanding model; when the understanding result is the understanding of the interpretation data, outputting the interpretation data as a query result.

[0096] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the data query method provided by the above-mentioned methods, the method comprising: receiving input data from a user, and organizing the input data; inputting the organized data into a professional knowledge model to obtain interpretation data output by the professional knowledge model, and inputting the interpretation data into a language understanding model to obtain an understanding result of the interpretation data output by the language understanding model; and when the understanding result is understanding of the interpretation data, outputting the interpretation data as a query result.

[0097] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0098] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0099] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. 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. However, 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 embodiments of the present invention.

Claims

1. A data query method, characterized in that: include: Receiving input data from a user, and collating the input data; Inputting the sorted data into a professional knowledge model to obtain interpretation data output by the professional knowledge model, and inputting the interpretation data into a language understanding model to obtain an understanding result of the interpretation data output by the language understanding model; When the understanding result is understanding of the explanation data, the explanation data is output as a query result.

2. The data query method according to claim 1, characterized in that: The step of arranging the input data includes: determining the type of the input data; If the type of the input data is a concept, then extracting the associated data of the concept from the knowledge base, and logically sorting the key data in the associated data; If the type of the input data is a book, the book is decomposed according to chapters to determine chapter data of each chapter, and the chapter data are logically sorted according to the chapters of the book; If the type of the input data is a paper, the paper is summarized in sections to determine summary data for each paragraph, and the summary data is logically sorted according to the paragraph information of the paper.

3. The data query method according to claim 2, characterized in that: The step of extracting the associated data of the concept from the knowledge base includes: Convert the concept into a vectorized representation to obtain a query vector; Performing preliminary indexing on the query vector to obtain preliminary index data; The preliminary index data is subjected to secondary indexing to obtain the associated data.

4. The data query method according to claim 2, characterized in that: The knowledge base is constructed based on: Extract preprint data from at least one data source; According to the type of the preprint data, vectorizing the preprint data; The knowledge base is constructed based on the vectorized preprint data.

5. The data query method according to claim 1, characterized in that: After inputting the sorted data into the professional knowledge model to obtain the interpretation data output by the professional knowledge model, and inputting the interpretation data into the language understanding model to obtain the understanding result of the interpretation data output by the language understanding model, the method further includes: When the understanding result is that the interpretation data is not understood, the understanding result is input into the professional knowledge model, so that the professional knowledge model performs iterative interpretation based on the understanding result until the interpretation data understood by the language understanding module is output.

6. The data query method according to claim 2, characterized in that: In the case where the understanding result is understanding the explanation data, after outputting the explanation data as the query result, the method further includes: Obtaining the scoring results of the explained data in multiple evaluation dimensions, as well as the user's correction suggestions for the explained data; The knowledge base and the knowledge interpretation algorithm are optimized according to the scoring results and the correction suggestions.

7. A data query device, characterized in that: include: A data sorting module, used for receiving input data from a user and sorting the input data; An interpretation module, used to input the sorted data into a professional knowledge model to obtain interpretation data output by the professional knowledge model, and input the interpretation data into a language understanding model to obtain an understanding result of the interpretation data output by the language understanding model; The output module is used to output the explanation data as a query result when the understanding result is understanding the explanation data.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the data query method according to any one of claims 1 to 6 is implemented.

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

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

Images