Scientific research assisting method and device based on generative model and computer equipment

Abstract

The invention relates to a scientific research auxiliary method and device based on a generative model and computer equipment, and is applied to the technical field of computer science. When the scientific research instruction comprises an experimental verification task, obtaining a knowledge base corresponding to the scientific research instruction based on an experimental verification agent, verifying the correctness of an experimental hypothesis in the experimental verification task based on the knowledge base, and generating a verification conclusion; wherein the knowledge base is established based on the pre-stored scientific research data, the problem that the experimental design in the scientific research task is lack of automatic analysis is solved, and the functions of combining the intelligent agent with the scientific research experiment and perfecting the scientific research assistance are realized.

Description

Technical Field

[0001] The present application relates to the field of computer science and technology, and in particular to a scientific research assistance method, device and computer equipment based on a generative model. Background Art

[0002] With the increasing amount of research data and literature in the scientific field, researchers are faced with huge workloads and challenges in acquiring information, organizing knowledge, designing experiments, and writing papers. Traditional literature management systems usually rely on keyword searches, which make it difficult to capture the deep semantic relationships between literature.

[0003] With the development of generative models (such as the GPT series) and multi-agent systems, the feasibility of automated processing tasks in scientific research has been significantly improved. However, existing systems based on generative models are mostly focused on natural language processing, text generation, and literature retrieval, and lack automated analysis of experimental design in scientific research tasks, so scientific research assistance is still insufficient in terms of comprehensiveness. Summary of the invention

[0004] In this embodiment, a scientific research assistance method, apparatus and computer equipment based on a generative model are provided to solve the problem of incomplete scientific research assistance in related technologies.

[0005] In a first aspect, a scientific research assistance method based on a generative model is provided in this embodiment, and the method includes: Obtain scientific research instructions submitted by users; When the scientific research instruction includes an experimental verification task, based on the experimental verification agent, a knowledge base corresponding to the scientific research instruction is obtained, and the correctness of the experimental hypothesis in the experimental verification task is verified based on the knowledge base to generate a verification conclusion; the knowledge base is established based on pre-stored scientific research data.

[0006] In some of the embodiments, before obtaining the knowledge base corresponding to the scientific research instruction, the method further includes: Extracting key information and core entities from the pre-stored scientific research materials based on the document management agent; Establish a knowledge graph based on the core entities; A knowledge base is established based on the key information and the knowledge graph.

[0007] In some of the embodiments, before obtaining the knowledge base corresponding to the scientific research instruction, the method further includes: Based on a knowledge discovery agent, performing knowledge reasoning on the knowledge graph in the knowledge base; The knowledge base is updated based on the result of the knowledge reasoning.

[0008] In some embodiments, when the scientific research instruction includes an experimental verification task, based on the experimental verification agent, a knowledge base corresponding to the scientific research instruction is obtained, and the correctness of the experimental hypothesis in the experimental verification task is verified based on the knowledge base, and a verification conclusion is generated, including: When the scientific research instruction includes an experimental verification task, based on the experimental verification agent, the experimental record information is obtained from the knowledge base, and the user-defined experimental design submitted by the user is obtained from the experimental verification task; Based on the semantic alignment result of the experiment record information and the custom experiment design, the consistency of the custom experiment design and the experiment record information is evaluated.

[0009] In some of the embodiments, after obtaining the user-submitted custom experimental design from the experimental verification task, the method further includes: Simulating the custom experimental design to obtain simulated experimental data; Obtaining real experimental data corresponding to the custom experimental design; Comparing the simulated experimental data with the real experimental data, analyzing the effectiveness of the custom experimental design, and generating an experimental improvement plan; The real experimental data and the experimental record information are classified and regressed, abnormal points in the custom experimental design are predicted, and experimental adjustment suggestions are generated based on the abnormal points.

[0010] In some embodiments, the method further comprises: When the scientific research instruction includes a search task, using crawler technology to crawl initial scientific research data matching the search task from a multi-source database; Based on the document management agent, the search information input by the user in the search task is obtained, and a query condition is generated based on the search information; the search information includes a subject, a keyword and a research intention; Based on the semantic similarity filtering algorithm, the scientific research data matching the query condition is filtered from the initial scientific research data; The scientific research data are stored and output.

[0011] In some embodiments, the method further comprises: When the scientific research instruction includes a knowledge reasoning task, the knowledge base is acquired based on a knowledge discovery agent and knowledge reasoning is performed on the knowledge graph in the knowledge base to obtain a reasoning result; Wherein, performing knowledge reasoning on the knowledge graph in the knowledge base to obtain reasoning results includes: Based on the text summarization algorithm and the knowledge graph in the knowledge base, a summary of the scientific research is generated to obtain a summary text; Based on rule reasoning, machine learning reasoning and the knowledge graph, the implicit relationship in the scientific research data is mined; based on the implicit relationship, a scientific hypothesis is obtained; Based on the path reasoning algorithm and the upstream and downstream relationships of the core entities in the knowledge graph, the research directions in the scientific research materials are analyzed to obtain research questions and experimental designs.

[0012] In some embodiments, the method further comprises: When the scientific research instruction includes a writing task, obtaining writing information in the knowledge base and the writing task based on a writing assistance agent, wherein the writing information includes a research topic and data; Generate a first draft of the article based on the knowledge base and the writing information; Conduct duplication detection and revision on the draft until the content of the article meets the originality requirements; Based on the content of the article, a reference record of the article is generated.

[0013] In a second aspect, a scientific research auxiliary device based on a generative model is provided in this embodiment, and the device includes: Instruction acquisition module, used to obtain scientific research instructions submitted by users; The result generation module is used to obtain the knowledge base corresponding to the scientific research instruction based on the experimental verification agent when the scientific research instruction includes an experimental verification task, verify the correctness of the experimental hypothesis in the experimental verification task based on the knowledge base, and generate a verification conclusion; the knowledge base is established based on pre-stored scientific research data.

[0014] In a third aspect, the present application further provides a computer device, wherein the computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the scientific research assistance method based on the generative model described in the first aspect is implemented.

[0015] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the generative model-based scientific research assistance method described in the first aspect is implemented.

[0016] Compared with the related art, the generative model-based scientific research assistance method, apparatus and computer equipment provided in this embodiment obtain scientific research instructions submitted by users; when the scientific research instructions include experimental verification tasks, based on the experimental verification agent, a knowledge base corresponding to the scientific research instructions is obtained, and the correctness of the experimental hypothesis in the experimental verification task is verified based on the knowledge base, and a verification conclusion is generated; wherein the knowledge base is established based on pre-stored scientific research data, which solves the problem of lack of automated analysis of experimental design in scientific research tasks, realizes the combination of intelligent agents and scientific research experiments, and improves the function of scientific research assistance.

[0017] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 The hardware structure block diagram of a terminal of a scientific research assistance method based on a generative model in one embodiment; Figure 2 1 is a flow chart of a scientific research assistance method based on a generative model in one embodiment; Figure 3 A schematic diagram of a user interaction interface of a scientific literature management system in one embodiment; Figure 4 is a schematic diagram of an intelligent agent in a scientific literature management system in one embodiment; Figure 5 A schematic diagram of a process flow of a scientific research assistance method based on a generative model in a preferred embodiment; Figure 6 It is a structural block diagram of a scientific research auxiliary device based on a generative model in one embodiment.

[0019] Figure numerals: 102, processor; 104, memory; 106, transmission device; 108, input and output device; 61, instruction acquisition module; 62, result generation module. DETAILED DESCRIPTION

[0020] In order to more clearly understand the purpose, technical solutions and advantages of the present application, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0021] Unless otherwise defined, the technical terms or scientific terms involved in this application shall have the general meaning understood by people with general skills in the technical field to which this application belongs. The words "one", "a", "the", "these" and the like in this application do not indicate a quantitative limitation, and they may be singular or plural. The terms "include", "comprise", "have" and any variants thereof involved in this application are intended to cover non-exclusive inclusions; for example, a process, method and system, product or device comprising a series of steps or modules (units) is not limited to the listed steps or modules (units), but may include unlisted steps or modules (units), or may include other steps or modules (units) inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there may be three relationships, for example, "A and / or B" may mean: A exists alone, A and B exist at the same time, and B exists alone. Generally, the character " / " indicates that the objects associated with each other are in an "or" relationship. The terms "first", "second", "third", etc. in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0022] The method embodiment provided in this embodiment can be executed in a terminal, a computer or a similar computing device. For example, running on a terminal, Figure 1 is a hardware structure block diagram of a terminal of the generative model-based scientific research assistance method of this embodiment. Figure 1 As shown, the terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 and a memory 104 for storing data, wherein the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA. The above terminal may also include a transmission device 106 and an input and output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is for illustration only and does not limit the structure of the above terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.

[0023] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the scientific research auxiliary method based on the generative model in this embodiment. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0024] The transmission device 106 is used to receive or send data via a network. The above network includes a wireless network provided by the communication provider of the terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (Radio Frequency, referred to as RF) module, which is used to communicate with the Internet wirelessly.

[0025] In this embodiment, a scientific research assistance method based on a generative model is provided. Figure 2 is a flow chart of the scientific research assistance method based on the generative model of this embodiment, such as Figure 2 As shown, the process includes the following steps: Step S210, obtaining the scientific research instructions submitted by the user.

[0026] Step S220, when the scientific research instruction includes an experimental verification task, based on the experimental verification agent, obtain the knowledge base corresponding to the scientific research instruction, verify the correctness of the experimental hypothesis in the experimental verification task based on the knowledge base, and generate a verification conclusion; the knowledge base is established based on pre-stored scientific research data.

[0027] Specifically, the method of this embodiment is applicable to a scientific document management system. Preferably, the scientific document management system may be a system for scientific research in a certain field, such as a life science document management system for the field of life science.

[0028] The system provides a generative interactive platform that can conduct dialogue and question-answering with users, and realize automated scientific research assistance, including literature retrieval, knowledge extraction, experimental verification, and writing assistance. Figure 3, the user enters the dialogue content related to scientific research needs in the input box of the user interaction interface (UI) of the system. Based on natural language processing (NLP) technology and large language model (LLM), the system parses the dialogue content submitted by the user and extracts scientific research instructions such as scientific research questions, experimental designs or literature keywords, and triggers the corresponding intelligent agent module to perform scientific research auxiliary work based on the scientific research instructions.

[0029] See also Figure 4 , the intelligent agent modules in the scientific literature management system include literature management agent (LMA), knowledge discovery agent (KDA), experimental verification agent (EVA) and writing assistance agent (WAA). Each intelligent agent (LMA, KDA, EVA, WAA) performs parallel computing through a distributed computing framework (such as Ray or Dask), and each intelligent agent communicates through RPC (remote procedure call) to ensure the real-time allocation and execution of tasks. Using a multi-agent collaboration strategy based on reinforcement learning (RL), the collaboration efficiency between agents is optimized through a reward mechanism (Reward Mechanism), ensuring that the system has efficient response capabilities when handling complex scientific research tasks. Using a multi-agent collaboration strategy based on reinforcement learning (RL), the collaboration efficiency between intelligent agents is optimized through a reward mechanism (Reward Mechanism), ensuring that the system has efficient response capabilities when handling complex scientific research tasks.

[0030] Scientific research materials include but are not limited to image materials, video materials, text materials, table materials, etc.

[0031] When the scientific research instructions include experimental verification tasks, the Experimental Verification Agent (EVA) compares the experimental hypothesis provided by the user with the existing scientific research data, verifies the experimental results, and provides improvement suggestions. The content of the scientific research data is stored in the form of a knowledge base, which can be established and updated based on the retrieval task in the previous dialogue, or updated based on the knowledge discovery task in the previous dialogue. The knowledge base includes a structured database and a graph database. The structured database is used to store standardized literature metadata and experimental data, and the graph database is used to store the relationship graph between experiments, supporting complex cross-literature association queries and discovering new knowledge.

[0032] For example, if a user is studying the expression of a gene in cancer, the system can automatically search for relevant literature, extract the experimental data and results involved, and build a knowledge base. Through knowledge base association analysis, the platform can generate a differential expression map of the gene in different cancer types. Subsequently, the user can define a new experimental hypothesis through an experimental verification agent, such as the expression changes of the gene under different drug treatment conditions. The experimental verification agent will provide experimental design suggestions and conduct a feasibility analysis of the plan based on existing experimental data.

[0033] Exemplarily, the user is studying the activity changes of a certain enzyme under different conditions, and the system can automatically search for relevant literature and extract the experimental data and results involved. Taking the study of the activity of a certain enzyme under different pH and temperature conditions as an example, the system will extract information such as experimental conditions, measurement methods, and result data from relevant literature. Subsequently, the system will automatically generate the activity change curve of the enzyme under different experimental environments based on the conditions and results of different experiments. For example, the user wants to study the activity changes of the enzyme in the presence of a certain inhibitor. Through the experimental verification agent, the user can define a new experimental hypothesis: whether the activity of the enzyme changes significantly under the action of different concentrations of inhibitors. The platform will automatically match the existing relevant data and literature, and generate experimental scheme suggestions based on these data. After the user submits the experiment, the platform can also verify the experimental data according to the results in the existing literature and generate an experimental result comparison report. In this embodiment, the platform extracts experimental conditions and enzyme activity data from multiple relevant documents through automatic literature crawling and information extraction technology, and stores them in a structured database. Through the graph database in the knowledge base, users can associate experimental data under different conditions across documents to further analyze the activity law of the enzyme. This data-driven literature management and experimental verification function greatly improves the data acquisition efficiency of scientific researchers under complex experimental conditions and provides a reliable experimental verification tool.

[0034] In this embodiment, scientific research instructions submitted by users are obtained; when the scientific research instructions include experimental verification tasks, based on the experimental verification agent, a knowledge base corresponding to the scientific research instructions is obtained, and the correctness of the experimental hypothesis in the experimental verification task is verified based on the knowledge base, and a verification conclusion is generated; wherein, the knowledge base is established based on pre-stored scientific research data, which solves the problem of lack of automated analysis of experimental designs in scientific research tasks, realizes the combination of intelligent agents and scientific research experiments, and improves the function of scientific research assistance.

[0035] In some of the embodiments, before obtaining the knowledge base corresponding to the scientific research instruction, the method further includes: Step S310, based on the document management agent, extract key information and core entities from pre-stored scientific research materials.

[0036] Step S320, establishing a knowledge graph based on core entities.

[0037] Step S330, establishing a knowledge base based on key information and knowledge graph.

[0038] Specifically, scientific research materials are obtained through user uploads or crawling by crawler systems. The document management agent indexes the obtained scientific research materials and stores them in Elasticsearch or other full-text search databases so that they can be quickly queried through the inverted index mechanism. The document management agent also extracts key information (such as authors, publication years, keywords, abstracts) and core entities (such as genes, proteins, and disease names) from the documents, and uses Neo4j to build a knowledge graph. The relationship between entities is generated through a relationship extraction algorithm, which includes BERT-based named entity recognition (NER) and dependency syntax analysis. The document management agent also includes topic model analysis based on latent Dirichlet allocation (LDA), automatically clusters the content of documents by topic, monitors the update of new documents in real time, and automatically generates research dynamic reports for users.

[0039] In some of the embodiments, before obtaining the knowledge base corresponding to the scientific research instruction, the method further includes: Step S410: Based on the knowledge discovery agent, knowledge reasoning is performed on the knowledge graph in the knowledge base.

[0040] Step S420: updating the knowledge base based on the result of knowledge reasoning.

[0041] Specifically, based on the graph enhancement generation technology in the knowledge discovery agent, we conduct in-depth analysis of scientific research data, explore the implicit relationship between data in the existing knowledge base, and generate scientific research hypotheses. The scientific research hypotheses are used to output to users for reference, and can also be used to update the knowledge base and enrich the existing knowledge base.

[0042] In a non-limiting embodiment, the scientific research instructions parsed from this round of dialogue include retrieval tasks, knowledge reasoning tasks, and experimental verification tasks, so the document management agent, knowledge discovery agent, and experimental verification agent will be triggered in sequence, realizing the entire process of document retrieval, knowledge base establishment and update, and experimental verification at one time, or in the scenario where the user wants to analyze the uploaded scientific research materials, the knowledge base is directly established based on the scientific research materials uploaded by the user, and experimental verification is performed. In another non-limiting embodiment, before the user provides the experimental verification task, the document management agent and the knowledge discovery agent have been triggered respectively through multiple rounds of dialogue, or only the document management agent has been triggered, so that the knowledge base has been established and saved. Therefore, in the current round of dialogue, the user only provides the experimental verification task in the dialogue, which can trigger the experimental verification agent and directly call the existing knowledge base to complete the corresponding scientific research auxiliary work.

[0043] In some embodiments, when the scientific research instruction includes an experimental verification task, based on the experimental verification agent, a knowledge base corresponding to the scientific research instruction is obtained, and the correctness of the experimental hypothesis in the experimental verification task is verified based on the knowledge base, and a verification conclusion is generated, including: Step S221, when the scientific research instruction includes an experimental verification task, based on the experimental verification agent, the experimental record information is obtained from the knowledge base, and the user-defined experimental design submitted by the user is obtained from the experimental verification task.

[0044] Step S222: based on the semantic alignment result of the experimental record information and the customized experimental design, the consistency between the customized experimental design and the experimental record information is evaluated.

[0045] Specifically, the Qi algorithm is based on the BERT embedding model and evaluates the consistency of the experimental design by measuring the similarity (cosinesimilarity) of the experimental process.

[0046] In some of the embodiments, after obtaining the user-submitted custom experimental design from the experimental verification task, the method further includes: Step S223, simulating the custom experimental design to obtain simulated experimental data.

[0047] Step S224, obtaining actual experimental data corresponding to the custom experimental design.

[0048] Step S225, compare the simulated experimental data with the real experimental data, analyze the effectiveness of the custom experimental design, and generate an experimental improvement plan.

[0049] Step S226, classify and regress the real experimental data and experimental record information, predict the abnormal points in the custom experimental design, and generate experimental adjustment suggestions based on the abnormal points.

[0050] Specifically, the experimental verification agent generates simulated experimental data by simulating experimental results through generative adversarial networks (GANs). The experimental verification agent uses ensemble learning algorithms (such as random forests and XGBoost) for classification and regression analysis to identify possible abnormal results in the experiment.

[0051] The consistency conclusions, experimental improvement plans, experimental adjustment suggestions and other verification conclusions generated by the experimental verification agent are presented to users through the user interface (UI) of the scientific literature management system. Users can selectively view the content of interest on the UI interface, or perform further data exploration by dragging or clicking.

[0052] In some of these embodiments, the method further comprises: Step S230, when the scientific research instruction includes a search task, crawler technology is used to capture initial scientific research data matching the search task from a multi-source database.

[0053] Specifically, adaptive web crawler technology is used to automatically crawl relevant literature data from multi-source life science databases (such as PubMed, ArXiv, BioRxiv, etc.), and the crawler system is adjusted in real time according to the crawled database structure.

[0054] Step S240, based on the document management agent, obtain the search information input by the user in the search task, and generate query conditions based on the search information; the search information includes topics, keywords and research intentions.

[0055] Step S250, based on the filtering algorithm of semantic similarity, filter the scientific research data matching the query conditions from the initial scientific research data; store and output the scientific research data.

[0056] Specifically, based on natural language processing technology and large language models, the user's input topics, keywords and research intentions are automatically parsed to generate query conditions for literature retrieval. The document filtering algorithm based on semantic similarity is used to automatically identify the documents that best match the query conditions. Furthermore, the matching documents are prioritized using the TF-IDF (term frequency-inverse document frequency) and BM25 algorithms, and the sorted documents are output to the questioner through the user interaction interface.

[0057] The pre-stored scientific research materials used in step S220 are obtained based on the search task. Before and after the experimental verification task, the user can trigger the document management agent by submitting a dialogue to obtain and update the pre-stored scientific research materials. Preferably, the scientific research materials are document materials obtained from a multi-source database.

[0058] In some of the embodiments, based on step S250, the scientific research data is stored, and thereafter, it also includes: based on the document management agent, extracting key information and core entities from the pre-stored scientific research data; establishing a knowledge graph based on the core entities; and establishing a knowledge base based on the key information and the knowledge graph.

[0059] Specifically, the retrieval task is used to instruct the document management agent to retrieve files and establish a knowledge base. If the current user has established a knowledge base through dialogue, the document management agent will regularly scan newly uploaded or crawled documents, automatically analyze the document content through a large-scale language model, identify the main content of the document, and extract experimental related information to update the knowledge base. The document management agent also has an adaptive analysis function, which can automatically adjust the extraction strategy according to the different sources of the document (journals, conferences, patents, etc.) and differences in document formats to ensure high-quality extraction results.

[0060] In some of these embodiments, the method further comprises: Step S260, when the scientific research instruction includes a knowledge reasoning task, the knowledge base is acquired based on the knowledge discovery agent and knowledge reasoning is performed on the knowledge graph in the knowledge base to obtain a reasoning result.

[0061] Specifically, the knowledge base supports users to conduct complex queries and association analysis, such as finding different results under common experimental conditions in different documents, or tracking the application of a specific experimental method in multiple fields. The knowledge discovery agent proposes new experimental hypotheses by combining existing literature and experimental data through technologies such as data mining, machine learning, and reasoning algorithms. Furthermore, users can verify the experimental hypotheses given by the knowledge discovery agent by triggering the experimental verification agent.

[0062] In some of the embodiments, knowledge reasoning is performed on the knowledge graph in the knowledge base to obtain reasoning results, including: Step S261, based on the text summarization algorithm and the knowledge graph in the knowledge base, generate a summary of the scientific research to obtain a summary text.

[0063] Step S262, based on rule reasoning, machine learning reasoning and knowledge graph, the implicit relationships in the scientific research data are mined; based on the implicit relationships, scientific hypotheses are obtained.

[0064] Step S263, based on the path reasoning algorithm and the upstream and downstream relationships of the core entities in the knowledge graph, analyze the research directions in the scientific research materials to obtain research questions and experimental designs.

[0065] Specifically, the knowledge discovery agent uses a text summarization algorithm based on a generative pre-trained model (such as GPT-4) to automatically generate abstracts and keywords for documents, and combines the generation of enhanced graph technology (GraphRAG) to embed the entity relationship graph in the document into the document text generation process, thereby improving the accuracy of knowledge reasoning. The knowledge discovery agent also uses the reasoning engine of the knowledge graph, combining rule reasoning and machine learning reasoning, to automatically mine implicit relationships in the document and derive new scientific hypotheses. The knowledge discovery agent supports knowledge retrieval functions based on path reasoning algorithms (such as Random Walk and TransE), which can infer potential research directions based on the upstream and downstream relationships of specific entities and recommend new research questions and experimental designs to users.

[0066] In some of these embodiments, the method further comprises: Step S270, when the scientific research instruction includes a writing task, based on the writing assistance agent, obtaining writing information in the knowledge base and the writing task, where the writing information includes research topics and data; Step S280, generating a first draft of the article based on the knowledge base and the writing information; performing duplication detection and modification on the first draft until the content of the article meets the originality requirements.

[0067] Step S290, generating a reference record of the article based on the content of the article.

[0068] Specifically, the writing assistant agent can generate a first draft of the paper based on the research topic and data input by the user. The first draft includes the introduction, methods, results and discussion sections. The writing assistant agent adopts the text generation technology based on the attention mechanism to ensure the logical consistency and scientific rationality of the generated content. Through the automatic reference recommendation module, the writing assistant agent can automatically match suitable references according to the written content, and use the BM25 algorithm to evaluate the semantic relevance between the literature and the writing content, give priority to recommending highly relevant literature, and automatically generate the reference format. The writing assistant agent has a built-in literature similarity detection module, and uses the Plagiarism Detection algorithm (such as SimHash or MinHash) to check the duplication of literature to ensure the originality of the paper content.

[0069] The present embodiment is described and illustrated below through preferred embodiments.

[0070] This preferred embodiment provides an automated platform ADMP-LS (Agent-Based Data Mining Platform for Life Sciences) for scientific research assistance. Figure 5 As shown in the figure, the platform workflow includes: S1. Literature import and data management: Users can upload new literature through the system or automatically crawl new literature through the platform's crawler system. The platform removes duplicates and formats the literature and stores it in an integrated literature management system. The literature management system supports query and screening based on multi-dimensional tags (such as research field, experimental method, key results, etc.), making it easy for users to quickly find relevant literature. This step can be implemented based on the above-mentioned literature management agent.

[0071] S2. Automatic document analysis and information extraction: The platform will regularly scan newly uploaded or crawled documents, automatically analyze the document content through a large-scale language model, identify the main content of the document, and extract experimental related information. The platform also has an adaptive analysis function, which can automatically adjust the extraction strategy according to the different sources of the document (journals, conferences, patents, etc.) and document formats to ensure high-quality extraction results. This step can be implemented based on the above-mentioned document management agent.

[0072] S3. Knowledge base update and management: After completing the literature information extraction, the system will import the structured data into the ES database and store the unstructured data and cross-literature associations in the Neo4j graph database. Through these databases, the platform supports users to perform complex queries and association analysis, such as finding different results under common experimental conditions in different literature, or tracking the application of a specific experimental method in multiple fields. This step can be implemented based on the above-mentioned literature management agent and knowledge discovery agent.

[0073] S4. Scientific research discovery and experimental verification: Users define research hypotheses or experimental questions. The platform will automatically match the stored relevant data and experimental results, and generate experimental scheme suggestions based on machine learning algorithms. The platform also supports automatic experimental result verification. Users can submit experimental data, and the platform will automatically compare the results in existing literature and provide a verification report. This step can be implemented based on the above-mentioned experimental verification agent.

[0074] In this preferred embodiment, through the multi-agent collaboration mechanism, the level and efficiency of scientific research assistance such as literature retrieval, knowledge discovery, and experimental verification are comprehensively improved. The data acquisition efficiency of researchers under complex experimental conditions is greatly improved, and a reliable experimental verification tool is provided.

[0075] It should be noted that the steps shown in the above process or the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0076] In this embodiment, a scientific research auxiliary device based on a generative model is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. The terms "module", "unit", "subunit", etc. used below can implement a combination of software and / or hardware of predetermined functions. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0077] Figure 6 is a structural block diagram of the scientific research auxiliary device based on the generative model of this embodiment, such as Figure 6 As shown, the device includes: an instruction acquisition module 61 and a result generation module 62.

[0078] The instruction acquisition module 61 is used to acquire the scientific research instructions submitted by the user.

[0079] The result generation module 62 is used to obtain the knowledge base corresponding to the scientific research instructions based on the experimental verification agent when the scientific research instructions include experimental verification tasks, verify the correctness of the experimental hypothesis in the experimental verification tasks based on the knowledge base, and generate a verification conclusion; the knowledge base is established based on pre-stored scientific research data.

[0080] In some of the embodiments, the result generation module 62 is also used to extract key information and core entities from pre-stored scientific research materials based on a document management agent; establish a knowledge graph based on core entities; and establish a knowledge base based on key information and the knowledge graph.

[0081] In some of the embodiments, the result generation module 62 is further used to perform knowledge reasoning on the knowledge graph in the knowledge base based on the knowledge discovery agent; and update the knowledge base based on the result of the knowledge reasoning.

[0082] In some of the embodiments, the result generation module 62 is also used to obtain experimental record information from the knowledge base based on the experimental verification agent and obtain the custom experimental design submitted by the user from the experimental verification task when the scientific research instructions include an experimental verification task; based on the semantic alignment results of the experimental record information and the custom experimental design, evaluate the consistency of the custom experimental design with the experimental record information.

[0083] In some of the embodiments, the result generation module 62 is also used to simulate the custom experimental design to obtain simulated experimental data; obtain real experimental data corresponding to the custom experimental design; Compare simulated experimental data with real experimental data, analyze the effectiveness of custom experimental design, and generate experimental improvement plans; classify and regress real experimental data and experimental record information, predict anomalies in custom experimental design, and generate experimental adjustment suggestions based on anomalies.

[0084] In some of the embodiments, the result generation module 62 is also used to, when the scientific research instructions include a retrieval task, use crawler technology to capture initial scientific research data that matches the retrieval task from a multi-source database; based on the document management agent, obtain the retrieval information input by the user in the retrieval task, and generate query conditions based on the retrieval information; the retrieval information includes topics, keywords and research intentions; based on the filtering algorithm of semantic similarity, filter the initial scientific research data to obtain scientific research data that matches the query conditions; and store and output the scientific research data.

[0085] In some of the embodiments, the result generation module 62 is also used to obtain the knowledge base based on the knowledge discovery agent and perform knowledge reasoning on the knowledge graph in the knowledge base to obtain reasoning results when the scientific research instructions include knowledge reasoning tasks; wherein, knowledge reasoning is performed on the knowledge graph in the knowledge base to obtain reasoning results, including: generating a summary of the scientific research based on the text summary algorithm and the knowledge graph in the knowledge base to obtain a summary text; mining implicit relationships in scientific research materials based on rule reasoning, machine learning reasoning and knowledge graphs; obtaining scientific hypotheses based on implicit relationships; analyzing the research directions in the scientific research materials based on the path reasoning algorithm and the upstream and downstream relationships of core entities in the knowledge graph to obtain research questions and experimental designs.

[0086] In some of the embodiments, the result generation module 62 is also used to obtain writing information in the knowledge base and the writing task based on the writing assistance agent when the scientific research instructions include a writing task, and the writing information includes the research topic and data; generate a first draft of the article based on the knowledge base and the writing information; perform duplication detection and modification on the first draft until the content of the article meets the originality requirements; and generate a reference record of the article based on the content of the article.

[0087] It should be noted that the above modules can be functional modules or program modules, and can be implemented by software or hardware. For modules implemented by hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0088] In this embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0089] Optionally, the computer device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0090] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and will not be repeated in this embodiment.

[0091] In addition, in combination with the generative model-based scientific research assistance method provided in the above embodiments, a storage medium can also be provided in this embodiment to implement the method. The storage medium stores a computer program; when the computer program is executed by the processor, any one of the generative model-based scientific research assistance methods in the above embodiments is implemented.

[0092] It should be understood that the specific embodiments described herein are only used to explain the application, rather than to limit it. Based on the embodiments provided in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of this application.

[0093] Obviously, the drawings are only some examples or embodiments of the present application. For ordinary technicians in the field, the present application can also be applied to other similar situations based on these drawings without creative work. In addition, it is understandable that although the work done in this development process may be complicated and lengthy, for ordinary technicians in the field, certain changes in design, manufacturing or production based on the technical content disclosed in this application are only conventional technical means and should not be regarded as insufficient content disclosed in this application.

[0094] The term "embodiment" in this application refers to a specific feature, structure or characteristic described in conjunction with the embodiment that can be included in at least one embodiment of the present application. The appearance of this phrase in various locations in the specification does not necessarily mean the same embodiment, nor does it mean that it is mutually exclusive with other embodiments and is independent or optional. It is clearly or implicitly understood by those of ordinary skill in the art that the embodiments described in this application can be combined with other embodiments without conflict.

[0095] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of patent protection. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the scope of protection of the present application. Therefore, the scope of protection of the present application shall be subject to the attached claims.

Claims

1. A scientific research assistance method based on a generative model, characterized in that: The method comprises: Obtain scientific research instructions submitted by users; When the scientific research instruction includes an experimental verification task, based on the experimental verification agent, a knowledge base corresponding to the scientific research instruction is obtained, and the correctness of the experimental hypothesis in the experimental verification task is verified based on the knowledge base to generate a verification conclusion; the knowledge base is established based on pre-stored scientific research data.

2. The scientific research assistance method based on the generative model according to claim 1 is characterized in that: Before obtaining the knowledge base corresponding to the scientific research instruction, it also includes: Extracting key information and core entities from the pre-stored scientific research materials based on the document management agent; Establish a knowledge graph based on the core entities; A knowledge base is established based on the key information and the knowledge graph.

3. The scientific research assistance method based on generative model according to claim 2 is characterized in that: Before obtaining the knowledge base corresponding to the scientific research instruction, it also includes: Based on a knowledge discovery agent, performing knowledge reasoning on the knowledge graph in the knowledge base; The knowledge base is updated based on the result of the knowledge reasoning.

4. The scientific research assistance method based on the generative model according to claim 2 or claim 3, characterized in that: When the scientific research instruction includes an experimental verification task, based on the experimental verification agent, a knowledge base corresponding to the scientific research instruction is obtained, and the correctness of the experimental hypothesis in the experimental verification task is verified based on the knowledge base to generate a verification conclusion, including: When the scientific research instruction includes an experimental verification task, based on the experimental verification agent, the experimental record information is obtained from the knowledge base, and the user-defined experimental design submitted by the user is obtained from the experimental verification task; Based on the semantic alignment result of the experiment record information and the custom experiment design, the consistency of the custom experiment design and the experiment record information is evaluated.

5. The scientific research assistance method based on generative model according to claim 4, characterized in that: After obtaining the user-submitted custom experimental design from the experimental verification task, it also includes: Simulating the custom experimental design to obtain simulated experimental data; Obtaining real experimental data corresponding to the custom experimental design; Comparing the simulated experimental data with the real experimental data, analyzing the effectiveness of the custom experimental design, and generating an experimental improvement plan; The real experimental data and the experimental record information are classified and regressed, abnormal points in the custom experimental design are predicted, and experimental adjustment suggestions are generated based on the abnormal points.

6. The scientific research assistance method based on generative model according to claim 1, characterized in that: The method further comprises: When the scientific research instruction includes a search task, using crawler technology to crawl initial scientific research data matching the search task from a multi-source database; Based on the document management agent, the search information input by the user in the search task is obtained, and a query condition is generated based on the search information; the search information includes a subject, a keyword and a research intention; Based on the semantic similarity filtering algorithm, the scientific research data matching the query condition is filtered from the initial scientific research data; The scientific research data are stored and output.

7. The scientific research assistance method based on generative model according to claim 2, characterized in that: The method further comprises: When the scientific research instruction includes a knowledge reasoning task, the knowledge base is acquired based on a knowledge discovery agent and knowledge reasoning is performed on the knowledge graph in the knowledge base to obtain a reasoning result; Wherein, performing knowledge reasoning on the knowledge graph in the knowledge base to obtain reasoning results includes: Based on the text summarization algorithm and the knowledge graph in the knowledge base, a summary of the scientific research is generated to obtain a summary text; Based on rule reasoning, machine learning reasoning and the knowledge graph, the implicit relationship in the scientific research data is mined; based on the implicit relationship, a scientific hypothesis is obtained; Based on the path reasoning algorithm and the upstream and downstream relationships of the core entities in the knowledge graph, the research directions in the scientific research materials are analyzed to obtain research questions and experimental designs.

8. The scientific research assistance method based on generative model according to claim 1, characterized in that: The method further comprises: When the scientific research instruction includes a writing task, obtaining writing information in the knowledge base and the writing task based on a writing assistance agent, wherein the writing information includes a research topic and data; Generate a first draft of the article based on the knowledge base and the writing information; Conduct duplication detection and revision on the draft until the content of the article meets the originality requirements; Based on the content of the article, a reference record of the article is generated.

9. A scientific research auxiliary device based on a generative model, characterized in that: The device comprises: Instruction acquisition module, used to obtain scientific research instructions submitted by users; The result generation module is used to obtain the knowledge base corresponding to the scientific research instruction based on the experimental verification agent when the scientific research instruction includes an experimental verification task, verify the correctness of the experimental hypothesis in the experimental verification task based on the knowledge base, and generate a verification conclusion; the knowledge base is established based on pre-stored scientific research data.

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

Images