Intelligent question and answer method and system
By combining algorithmic control with an intelligent source Turing machine based on a set of intelligent sources, the problem of insufficient understanding of user intent in intelligent interaction in traditional office software is solved, enabling fast and accurate responses and personalized suggestions, thus improving the user experience.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-30
- Publication Date
- 2026-04-10
AI Technical Summary
Traditional office software lacks a deep understanding of user intent in terms of intelligent interaction, and cannot provide personalized suggestions or automated decision support, resulting in a fragmented user experience, which is prone to misunderstanding, especially when faced with complex queries.
The intelligent source Turing machine, which combines algorithmic control and an intelligent source set, processes user questions through semantic search, preprocessing, and various machine learning models to generate fast and accurate response information, including highlighting PDF text and providing relevant question suggestions.
It enables quick and accurate answers to user questions in complex office scenarios, improving user experience and providing personalized suggestions and automated decision support.
Smart Images
Figure CN121833897A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of computer technology, and in particular to an intelligent question-answering method and system. Background Technology
[0002] Traditional office software (such as Word and WPS) has basic document processing capabilities, but it has significant limitations in intelligent interaction. It lacks a deep understanding of user intent and cannot provide personalized suggestions or automated decision support, resulting in a fragmented user experience.
[0003] With the development of AI technology, large models (such as GPT and DeepSeek) have provided new paths to overcome these bottlenecks due to their powerful natural language processing capabilities. Currently, many office software programs have integrated large models, but they are only connected to a specific large language model (LLM). For example, Word's Co-Pilot is connected to ChatGPT as its LLM.
[0004] If only one large model is used, there will be functional limitations and user experience bottlenecks in complex office scenarios. For example, when faced with vague or complex queries, misunderstandings or irrelevant answers are likely to occur, affecting the user experience. Summary of the Invention
[0005] In view of this, the present invention provides an intelligent question-answering method and system, which, by combining algorithmic control and an intelligent source Turing machine of intelligent source set, can quickly and accurately answer users' questions, greatly improving the user experience.
[0006] The first aspect of this invention provides an intelligent question-answering method, comprising:
[0007] Receiving user questions;
[0008] A semantic search is performed based on the user's question to obtain semantic search results;
[0009] The semantic search results and the PDF text collected by the system are preprocessed to obtain the target data;
[0010] The target data and the user question are input into an intelligent source Turing machine, which outputs a response. The intelligent source Turing machine includes an algorithm control device and an intelligent source set. The algorithm control device includes multiple pre-trained machine learning models, and the intelligent source set includes multiple large language models. The target data is decomposed into multiple sub-tasks using the algorithm control device, and a processing method is determined for each sub-task. The sub-tasks are processed according to their respective processing methods to obtain processing results. A response is generated based on all processing results. The processing method is either machine learning model processing or large language model processing, and the response includes the answer to the user question.
[0011] Optionally, the response information may also include related questions about the user's question.
[0012] Optionally, after inputting the target data and the user question into the intelligent source Turing machine and outputting the response information, the method further includes:
[0013] The relevant portions of the response information are highlighted in the PDF text.
[0014] Optionally, after preprocessing the semantic search results and the PDF text collected by the system to obtain the target data, the method further includes:
[0015] The target data is stored in the database.
[0016] A second aspect of the present invention provides an intelligent question-answering system, comprising:
[0017] The receiving unit is used to receive user questions;
[0018] A semantic search unit is used to perform a semantic search based on the user's question to obtain semantic search results.
[0019] The preprocessing unit is used to preprocess the semantic search results and the PDF text collected by the system to obtain target data;
[0020] An intelligent response unit is used to input the target data and the user question into an intelligent source Turing machine and output response information. The intelligent source Turing machine includes an algorithm control device and an intelligent source set. The algorithm control device includes multiple pre-trained machine learning models, and the intelligent source set includes multiple large language models. The target data is decomposed into multiple sub-tasks using the algorithm control device, and a processing method is determined for each sub-task. The sub-tasks are processed according to their respective processing methods to obtain processing results. Response information is generated based on all processing results. The processing method is either machine learning model processing or large language model processing, and the response information includes the answer to the user question.
[0021] Optionally, the response information may also include related questions about the user's question.
[0022] Optionally, the intelligent question-answering system further includes:
[0023] A highlighting unit is used to highlight the relevant portions of the response information in the PDF text.
[0024] Optionally, the intelligent question-answering system further includes:
[0025] A storage unit is used to store the target data into a database.
[0026] As can be seen from the above scheme, the present invention provides an intelligent question-answering method and system. After obtaining semantic search results by performing semantic search on the user's question, the semantic search results and the PDF text collected by the system are preprocessed to obtain target data. Finally, the target data is processed by an intelligent source Turing machine that combines algorithm control and intelligent source set to obtain answer information for the user's question. This achieves the purpose of quickly and accurately answering the user's question and greatly improves the user experience. Attached Figure Description
[0027] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0028] Figure 1 A detailed flowchart of an intelligent question-answering method provided in an embodiment of the present invention;
[0029] Figure 2 A schematic diagram of a question-answering architecture is provided for another embodiment of the present invention;
[0030] Figure 3 A schematic diagram of an intelligent question-and-answer interface provided for another embodiment of the present invention;
[0031] Figure 4 A schematic diagram of an intelligent question-and-answer interface provided for another embodiment of the present invention;
[0032] Figure 5 This is a schematic diagram of an intelligent question-answering system provided in another embodiment of the present invention. Detailed Implementation
[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0034] The term "comprising" and its variations as used herein are open-ended inclusions, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below.
[0035] It should be noted that the information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this invention are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.
[0036] It should be noted that the concepts of "first" and "second" mentioned in this invention are only used to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.
[0037] It should be noted that the terms "a" and "a plurality of" used in this invention are illustrative rather than restrictive. Those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".
[0038] This invention provides an intelligent question-answering method, such as... Figure 1 As shown, the specific steps include:
[0039] S101, User Receiving Issues.
[0040] In the practical application of this invention, a question and answer area can be provided for users to ask questions and for the questions to be answered.
[0041] S102. Perform semantic search based on user questions to obtain semantic search results.
[0042] Semantic search, by understanding user intent and contextual relationships, can more accurately match complex needs. It can use, but is not limited to, Natural Language Processing (NLP) techniques to analyze word meaning, grammar, and contextual relationships, such as Latent Semantic Indexing (LSI).
[0043] Specifically, after performing semantic search based on user questions to obtain user intent and contextual relationships, a database is generated by semantically searching PDF documents crawled from trusted sources, and appropriate materials are selected from it as semantic search results.
[0044] S103. Preprocess the semantic search results and the PDF text collected by the system to obtain the target data.
[0045] The PDF texts collected by the system include, but are not limited to, PDF texts uploaded by users.
[0046] In the actual application of this invention, the preprocessing methods include, but are not limited to, extracting the main text content, determining the position of the sentence on the page, inserting dynamic placeholders, calculating hash values, and detecting duplicate articles, etc., which are not limited here.
[0047] Optionally, in another embodiment of the present invention, after inputting the target data and user question into the intelligent source Turing machine and outputting the obtained response information, the method further includes:
[0048] Store the target data in the database.
[0049] The database includes user private libraries, records of user queries and answers, shared libraries, etc., which are not limited here.
[0050] S104. Input the target data and user question into the intelligent source Turing machine and output the response information.
[0051] The intelligent source Turing machine includes an algorithm control device and an intelligent source set. The algorithm control device includes multiple pre-trained machine learning models, and the intelligent source set includes multiple large language models. The algorithm control device decomposes the target data into multiple sub-tasks and determines the processing method corresponding to each sub-task. The sub-tasks are processed according to the processing methods corresponding to the sub-tasks to obtain processing results. Response information is generated based on all processing results. The processing method is either using machine learning models or large language models, and the response information includes the answer to the user's question.
[0052] It should be noted that pre-trained machine learning models include deep neural networks, traditional machine learning models, etc., which are not limited here; the intelligent source set includes paid large language models, internally trained large language models, etc., which are not limited here.
[0053] It should be noted that the pre-trained machine learning model is a machine learning model trained based on writing scenario data, while the internally trained large language model is a large language model that has been fine-tuned based on writing scenario data, and no restrictions are imposed here.
[0054] This invention trains a machine learning model based on writing scenario data and fine-tunes a large language model, enabling the generated pre-trained machine learning model and the internally trained large language model to more effectively analyze and answer user questions.
[0055] In the practical application of this invention, the processing method of the subtask is determined according to different subtask types. For example, the content containing semantic parts of the problem is processed using LLM, the content containing reasoning and mathematics is processed using Large Reasoning Model (LRM), and the part that requires images is processed using Large Vision Model (LVM).
[0056] Optionally, in another embodiment of the present invention, the algorithm control device further includes: a task decomposition and integration module, an NLP algorithm cluster, a text engineering algorithm cluster, a text mining algorithm cluster, and a combinatorial optimization algorithm cluster.
[0057] Text engineering and text mining are two core directions in the field of Natural Language Processing (NLP). The former focuses on technical implementation, while the latter focuses on knowledge discovery. Text engineering builds the infrastructure for text processing, mainly referring to the engineering methods that provide technical support for text analysis and processing, including text representation, feature extraction, and model optimization. It involves transforming raw text into structured data that computers can process, providing underlying support for upper-level applications. Text mining extracts valuable knowledge from text, leaning more towards the application level. It aims to extract novel, useful, and understandable patterns or knowledge from unstructured text data, using machine learning, NLP, and other technologies to analyze the text, thereby decomposing information into tasks. A cluster of combinatorial optimization algorithms is used to analyze each subtask and determine which processing method is more suitable for each subtask.
[0058] In the practical application of this invention, the processing method can be selected based on the advantages and disadvantages of various intelligent sources studied in advance and the needs of the current task.
[0059] Optionally, in another embodiment of the invention, the response information may also include related questions about the user's question.
[0060] In the practical application of this invention, the relevant questions for user questions can be heuristic suggestions. For example, based on the uploaded article content and the questions raised by the user, the intelligent source Turing machine infers the user's interests and infers related questions accordingly. Based on the degree of relevance to the questions already raised, a relevant question that the user has not asked is selected and returned to the user to stimulate their thinking.
[0061] Optionally, in another embodiment of the present invention, after inputting the target data and user question into the intelligent source Turing machine and outputting the obtained response information, the method further includes:
[0062] Highlight the relevant parts of the response information in the PDF text.
[0063] like Figure 2 The diagram illustrates a question-answering architecture provided by an embodiment of the present invention. It can be applied to, but is not limited to, the Doenba system. The architecture includes a preprocessing section, a preprocessed database of academic materials and papers, an algorithm control device, an intelligent source, and a database of PDF documents crawled from trusted sources after semantic search processing. Upon receiving a user's question about an input document, the system performs a semantic search on the user's input topic and selects suitable materials from the semantically searched PDF documents crawled from trusted sources as semantic search results. The semantic search results and the user-uploaded PDF text are preprocessed to obtain target data, which is then input into the algorithm control device. The intelligent source Turing machine (algorithm control device + intelligent source) combines the target data, the user's question about the input PDF document, the preprocessed database of academic materials and papers, and the database of semantically searched PDF documents crawled from trusted sources to process the target data, obtaining answer information for the user's question, highlighting relevant parts of the PDF document, and providing heuristic suggestions.
[0064] like Figure 3 and Figure 4 As shown, users may have questions about certain basic concepts or want to ask about LaTeX typesetting during the writing process. They can get answers directly through the intelligent question-answering method (Message Doenba) provided by this invention. No limitation is made here.
[0065] As can be seen from the above scheme, the present invention provides an intelligent question answering method. After obtaining semantic search results by performing semantic search on user questions, the semantic search results and PDF text collected by the system are preprocessed to obtain target data. Finally, the target data is processed by an intelligent source Turing machine that combines algorithm control and intelligent source set to obtain answer information for user questions. This achieves the goal of quickly and accurately answering user questions and greatly improves user experience.
[0066] Another embodiment of the present invention provides an intelligent question-answering system, such as... Figure 5 As shown, it specifically includes:
[0067] The receiving unit 501 is used to receive user questions.
[0068] In the practical application of this invention, a question and answer area can be provided for users to ask questions and for the questions to be answered.
[0069] Semantic search unit 502 is used to perform semantic search based on user questions and obtain semantic search results.
[0070] Semantic search, by understanding user intent and contextual relationships, can more accurately match complex needs. It can employ, but is not limited to, natural language processing (NLP) techniques to analyze word meaning, syntax, and contextual relationships, such as Latent Semantic Index (LSI).
[0071] Specifically, after performing semantic search based on user questions to obtain user intent and contextual relationships, a database is generated by semantically searching PDF documents crawled from trusted sources, and appropriate materials are selected from it as semantic search results.
[0072] The preprocessing unit 503 is used to preprocess the semantic search results and the PDF text collected by the system to obtain the target data.
[0073] The PDF texts collected by the system include, but are not limited to, PDF texts uploaded by users.
[0074] In the actual application of this invention, the preprocessing methods include, but are not limited to, extracting the main text content, determining the position of the sentence on the page, inserting dynamic placeholders, calculating hash values, and detecting duplicate articles, etc., which are not limited here.
[0075] The intelligent response unit 504 is used to input the target data and user questions into the intelligent source Turing machine and output the response information.
[0076] The intelligent source Turing machine includes an algorithm control device and an intelligent source set. The algorithm control device includes multiple pre-trained machine learning models, and the intelligent source set includes multiple large language models. The algorithm control device decomposes the target data into multiple sub-tasks and determines the processing method corresponding to each sub-task. The sub-tasks are processed according to the processing methods corresponding to the sub-tasks to obtain processing results. Response information is generated based on all processing results. The processing method is either using machine learning models or large language models, and the response information includes the answer to the user's question.
[0077] It should be noted that pre-trained machine learning models include deep neural networks, traditional machine learning models, etc., which are not limited here; the intelligent source set includes paid large language models, internally trained large language models, etc., which are not limited here.
[0078] It should be noted that the pre-trained machine learning model is a machine learning model trained based on writing scenario data, while the internally trained large language model is a large language model that has been fine-tuned based on writing scenario data, and no restrictions are imposed here.
[0079] This invention trains a machine learning model based on writing scenario data and fine-tunes a large language model, enabling the generated pre-trained machine learning model and the internally trained large language model to more effectively analyze and answer user questions.
[0080] In the practical application of this invention, the processing method of the subtask is determined according to different subtask types. For example, LLM is used for the semantic part of the problem, LRM is used for the reasoning and mathematical part, and LVM is used for the part that requires images.
[0081] For details on the specific operation of the units disclosed in the above embodiments of the present invention, please refer to the corresponding method embodiments, such as... Figure 1 As shown, it will not be elaborated further here.
[0082] Optionally, in another embodiment of the invention, the response information may also include related questions about the user's question.
[0083] The specific working process of the units disclosed in the above embodiments of the present invention can be found in the corresponding method embodiments, and will not be repeated here.
[0084] Optionally, in another embodiment of the present invention, one implementation of the intelligent question-answering system further includes:
[0085] The highlighting unit is used to highlight relevant parts of the response information in the PDF text.
[0086] The specific working process of the units disclosed in the above embodiments of the present invention can be found in the corresponding method embodiments, and will not be repeated here.
[0087] Optionally, in another embodiment of the present invention, one implementation of the intelligent question-answering system further includes:
[0088] Storage unit, used to store target data into the database.
[0089] The specific working process of the units disclosed in the above embodiments of the present invention can be found in the corresponding method embodiments, and will not be repeated here.
[0090] As can be seen from the above scheme, the present invention provides an intelligent question-answering system. After obtaining semantic search results by performing semantic search on user questions, the semantic search results and PDF text collected by the system are preprocessed to obtain target data. Finally, the target data is processed by an intelligent source Turing machine that combines algorithm control and intelligent source set to obtain answer information for user questions. This achieves the goal of quickly and accurately answering user questions and greatly improves user experience.
[0091] Another embodiment of the present invention provides a computer program product, which, when executed, is used to perform the above-described intelligent question-answering method.
[0092] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from a ROM. When the computer program is executed by a processing device, it performs the functions defined in the methods of the embodiments of the present invention.
[0093] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in this invention is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely exemplary forms for implementing the invention.
[0094] While several specific implementation details are included in the foregoing discussion, these should not be construed as limiting the scope of the invention. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.
[0095] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention is not limited to the specific combination of the above-described technical features, but also includes other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above-described features with, but not limited to, technical features with similar functions in the present invention.
Claims
1. An intelligent question-answering method, characterized in that, include: Receiving user questions; A semantic search is performed based on the user's question to obtain semantic search results; The semantic search results and the PDF text collected by the system are preprocessed to obtain the target data; The target data and the user question are input into an intelligent source Turing machine, which outputs a response. The intelligent source Turing machine includes an algorithm control device and an intelligent source set. The algorithm control device includes multiple pre-trained machine learning models, and the intelligent source set includes multiple large language models. The target data is decomposed into multiple sub-tasks using the algorithm control device, and a processing method is determined for each sub-task. The sub-tasks are processed according to their respective processing methods to obtain processing results. A response is generated based on all processing results. The processing method is either machine learning model processing or large language model processing, and the response includes the answer to the user question.
2. The intelligent question-answering method according to claim 1, characterized in that, The response information also includes related questions about the user's question.
3. The intelligent question-answering method according to claim 1, characterized in that, After inputting the target data and the user question into the intelligent source Turing machine and outputting the response information, the process further includes: The relevant portions of the response information are highlighted in the PDF text.
4. The intelligent question-answering method according to claim 1, characterized in that, After preprocessing the semantic search results and the PDF text collected by the system to obtain the target data, the process further includes: The target data is stored in the database.
5. An intelligent question-answering system, characterized in that, include: The receiving unit is used to receive user questions; A semantic search unit is used to perform a semantic search based on the user's question to obtain semantic search results. The preprocessing unit is used to preprocess the semantic search results and the PDF text collected by the system to obtain target data; An intelligent response unit is used to input the target data and the user question into an intelligent source Turing machine and output response information. The intelligent source Turing machine includes an algorithm control device and an intelligent source set. The algorithm control device includes multiple pre-trained machine learning models, and the intelligent source set includes multiple large language models. The target data is decomposed into multiple sub-tasks using the algorithm control device, and a processing method is determined for each sub-task. The sub-tasks are processed according to their respective processing methods to obtain processing results. Response information is generated based on all processing results. The processing method is either machine learning model processing or large language model processing, and the response information includes the answer to the user question.
6. The intelligent question-answering system according to claim 5, characterized in that, The response information also includes related questions about the user's question.
7. The intelligent question-answering system according to claim 5, characterized in that, Also includes: A highlighting unit is used to highlight the relevant portions of the response information in the PDF text.
8. The intelligent question-answering system according to claim 5, characterized in that, Also includes: A storage unit is used to store the target data into a database.