Scientific researcher and scientific research project matching recommendation method and system

Abstract

The invention belongs to the technical field of feature matching recommendation, and discloses a scientific researcher and scientific research project matching recommendation method and system, and the method comprises the steps: collecting and processing scientific researcher information, and generating a structured summary and evaluation through semantic slicing, vectorization and an intelligent agent; based on this, a scientific researcher portrait including a public portrait and a dynamic academic ability portrait is generated; scientific research project information is collected and processed, and an AI intelligent agent is utilized to analyze project documents to generate project features and hierarchical demands; and executing matching recommendation, including multi-stage single-person matching based on rules, academic requirements and project experience requirements, and performing global evaluation and comparative sorting on passing persons. The system comprises a corresponding scientific research personnel information processing module, a scientific research personnel portrait generation module, a scientific research project information processing module and a matching recommendation module. According to the invention, matching recommendation can be accurately and efficiently performed for scientific researchers and scientific research projects.

Description

Technical Field

[0001] The present application belongs to the field of feature matching recommendation technology, and specifically relates to a matching recommendation method and system for scientific researchers and scientific research projects. Background Art

[0002] In current scientific research management practices, with the explosive growth of scientific research data and the increasing complexity and interdisciplinary nature of research projects, traditional management models face severe challenges. Assessing researchers' competence typically relies on manual collection, organization, and analysis of information such as their academic achievements (such as papers and patents), project experience, and educational background. This process is not only time-consuming and labor-intensive, but also prone to missing key information, resulting in incomplete and inaccurate assessments. Furthermore, evaluation criteria are often subjective, and different evaluators may reach widely divergent conclusions, impacting the fairness and scientific nature of talent selection and utilization.

[0003] On the other hand, for scientific research projects, especially complex interdisciplinary ones, quickly and accurately identifying the most suitable participants or leaders from among a large number of researchers is crucial to project success. Existing technologies have limited means of extracting and analyzing project information, making it difficult to quickly grasp a project's core requirements (such as key technologies, research directions, and experience requirements). This often leads to a poor match between talent and projects due to insufficient understanding of personnel capabilities or information asymmetry when forming research teams, impacting project progress and quality, and even leading to waste of resources. Furthermore, an incomplete understanding of researchers' comprehensive capabilities and potential can also lead to irrational allocation of academic resources (such as funding, equipment, and opportunities), impacting the output and transformation efficiency of scientific research results.

[0004] Therefore, existing technologies have obvious deficiencies in the efficiency of scientific researchers' information acquisition, objectivity of evaluation, accuracy of person-job matching, and rationality of resource allocation. There is an urgent need for a more accurate and efficient matching recommendation method. Summary of the Invention

[0005] The present invention aims to overcome the shortcomings of the prior art, such as difficulty in obtaining information about scientific researchers, cumbersome and subjective evaluation, low efficiency in talent-project matching, and irrational resource allocation, and to provide a method and system that can accurately and efficiently match scientific researchers with scientific research projects.

[0006] In order to solve the above technical problems, this application is implemented as follows: In a first aspect, an embodiment of the present application provides a method for recommending matching of scientific researchers and scientific research projects, the method comprising: Researcher information collection and processing steps: collecting structured basic information of researchers; receiving and processing unstructured documents related to the researchers, said processing including semantic slicing and vectorizing the document content, generating structured document summary data using the first type of intelligent agent, and evaluating the summary data using the second type of intelligent agent to generate evaluation data; A researcher portrait generation step: based on the structured basic information, the processed unstructured documents, the summary data, and the evaluation data, a third-type intelligent agent is used to generate a researcher portrait, wherein the researcher portrait includes a public portrait and / or an academic ability portrait for a specific ability query; Research project information collection and analysis steps: Collect basic information about the research project; receive and process project documents related to the research project, use the fourth type of agent to analyze the content of the project documents and the basic project information, and generate structured project feature data and hierarchical project requirements, which include at least basic requirements, academic requirements, and project experience requirements; Matching recommendation steps: receive the scientific research project to be matched and at least one scientific researcher to be matched; perform the single-person matching stage for each scientific researcher to be matched, and the matching stage includes at least basic rule matching based on the basic requirements of the scientific research project to be matched and the basic information of the scientific researcher to be matched, academic requirements matching based on the academic requirements of the project and the scientific researcher portrait and related document summary, and project experience requirement matching based on the project experience requirements of the scientific research project to be matched and the scientific researcher portrait; perform the multi-person comparison and ranking stage for the scientific researchers who pass the single-person matching stage to generate recommended ranking results.

[0007] As an optional implementation of the first aspect of the present application, in the scientific researcher information collection and processing step, the step of receiving and processing unstructured documents related to the scientific researcher also includes: performing enhancement processing on a specific type of unstructured document, and the enhancement processing includes at least one or a combination of the following operations: obtaining supplementary information related to a specific type of unstructured document from an external academic database or information source through a web crawler or API interface, and the supplementary information includes at least one of the journal name, journal division, impact factor, author list or project funding information in which the specific type of document is published; determining the author level based on the signature information of the author in the specific type of unstructured document input by the user or identified by the system; and performing quality grading or labeling on the specific type of unstructured document based on the supplementary information and / or the author level and according to preset rules.

[0008] As an optional implementation of the first aspect of the present application, in the scientific researcher portrait generation step, generating the academic ability portrait for the specific ability query specifically includes: receiving keywords describing specific academic abilities or research directions input by the user; calling the keyword extraction agent, processing the user input, extracting query keywords, and generating an extended keyword set containing relevant terms; using the query keywords and / or the extended keyword set as the retrieval basis, calling the screening and sorting module, the screening and sorting module adopts a hybrid retrieval technology, searching in the structured document summary data of the scientific researcher, screening out relevant document summary data, and forming a relevant summary data subset; calling the dimension generation agent, generating an evaluation dimension for evaluating the specific academic ability according to the query intention input by the user; calling the academic ability portrait generation agent, generating an academic ability portrait report for the specific query ability based on the relevant summary data subset and the evaluation dimension.

[0009] As an optional implementation of the first aspect of the present application, the screening and sorting module adopts a hybrid retrieval technology, and its specific implementation includes at least: performing full-text retrieval: using the database to establish an index for the preset text field in the structured document summary data, finding and recalling the document summary records containing the query keyword; and performing semantic retrieval: converting the query keyword or query statement into a query vector, calculating the similarity between the query vector and the feature vector of the document summary data stored in the vector database, and retrieving semantically related document summary records; fusing or sorting the full-text retrieval results and the semantic retrieval results, and outputting a subset of relevant summary data.

[0010] As an optional implementation of the first aspect of the present application, the step of generating a portrait of a scientific researcher also includes generating a brief description of achievements, and the specific implementation includes: classifying unstructured documents related to scientific researchers into project and achievement information based on document content, metadata or user-specified type tags; storing the classified project and achievement information in a structured manner in a database, and performing update and merge operations; when the need to generate a brief description of achievements is triggered, extracting core field information from the project and achievement information under the name of the scientific researcher from the database to form a structured information list; calling a large language model, taking the structured information list as input, and generating a natural language text summary of the scientific researcher's project experience and academic achievements based on preset prompts.

[0011] As an optional implementation method of the first aspect of the present application, in the single-person matching stage in the matching recommendation step, before executing the academic needs matching and the project experience needs matching, it also includes executing an information preparation sub-step, which at least includes: using the project feature data or project needs analysis results of the scientific research project to be matched as query input; searching and sorting the personal knowledge base of the current scientific researcher to be matched, and screening out a subset of document summary data related to the scientific research project to be matched; judging whether the scientific researcher to be matched already has an academic ability portrait related to the ability required for the project to be matched, and if not, generating an academic ability portrait for preparation based on the keywords of the project to be matched; and using the screened out relevant summary subset and the relevant portrait for preparation as input for subsequent matching steps.

[0012] As an optional implementation method of the first aspect of the present application, the multi-person comparison and ranking stage of the matching recommendation step specifically includes: executing a global evaluation sub-step: for each candidate scientific researcher, extracting the key academic achievements and project experience indicators preset in the candidate knowledge base; calculating the global capability evaluation score based on the extracted indicators according to the preset evaluation model; executing a comprehensive ranking sub-step: comprehensively considering the academic needs matching score, project experience needs matching score and calculated global capability evaluation score obtained by each candidate scientific researcher in the single matching stage, and calculating the comprehensive matching total score through a preset algorithm; and ranking all candidates according to the comprehensive matching total score.

[0013] In a second aspect, an embodiment of the present application provides a matching recommendation system for scientific researchers and scientific research projects, the system comprising: A researcher information processing module is used to collect structured basic information about researchers; receive and process unstructured documents related to the researchers, including semantic slicing and vectorizing the document content, generating structured document summary data using the first type of intelligent agent, and evaluating the summary data using the second type of intelligent agent to generate evaluation data; A researcher portrait generation module is configured to generate a researcher portrait using a third type of agent based on the structured basic information, the processed unstructured documents, the summary data, and the evaluation data. The researcher portrait includes a public portrait and / or an academic ability portrait for a specific ability query. The scientific research project information processing module is used to collect basic information about scientific research projects; receive and process project documents related to the scientific research projects, analyze the content of the project documents and basic project information using the fourth type of agent, and generate structured project feature data and hierarchical project requirements, which include at least basic requirements, academic requirements, and project experience requirements; The matching recommendation module is used to receive scientific research projects to be matched and at least one scientific researcher to be matched; a single matching stage is performed for each scientific researcher to be matched, and the matching stage at least includes basic rule matching based on the basic requirements of the scientific research project to be matched and the basic information of the scientific researcher to be matched, academic requirements matching based on the academic requirements of the project and the scientific researcher portrait and related document summary, and project experience requirement matching based on the project experience requirements of the scientific research project to be matched and the scientific researcher portrait; a multi-person comparison and ranking stage is performed for the scientific researchers who pass the single matching stage to generate a recommended ranking result.

[0014] In a third aspect, an embodiment of the present application provides an electronic device comprising a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein when the program or instruction is executed by the processor, the steps of the method described in the first aspect are implemented.

[0015] In a fourth aspect, an embodiment of the present application provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented.

[0016] Compared with the prior art, the present invention proposes a matching recommendation method for scientific researchers and scientific research projects, which has the following beneficial effects: 1. Accurate and efficient talent matching: Through deep semantic understanding (vectorization and agent-based analysis) of researchers and project documents and the comprehensive utilization of multi-dimensional information (basic information, papers, project experience, and external data), combined with dynamically generated profiles and hierarchical project requirements, we achieve more accurate talent-project matching than traditional manual screening or simple keyword matching. A phased matching strategy improves efficiency and reduces unnecessary calculations.

[0017] 2. Comprehensive and objective character portrait construction: Automatically integrate multi-source, heterogeneous data (structured and unstructured), leveraging AI-powered large models to generate structured, multi-dimensional profiles of researchers, overcoming the subjectivity and incompleteness of manual assessments. Profile content can be dynamically generated based on query requirements, making it more targeted and timely.

[0018] 3. Scientific and rational decision support: Based on data-driven analysis and quantitative evaluation results, it provides scientific and transparent decision-making basis for scientific research managers in team formation, project allocation, resource allocation, etc. The generated matching report includes detailed reasons, enhancing the interpretability of the results.

[0019] 4. High degree of automation in information processing: This significantly reduces the manual workload of collecting, organizing, reading, and analyzing scientific research personnel and project data, improving the overall efficiency of scientific research management. Structured and vectorized processing facilitates the storage, retrieval, and reuse of information. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a flow chart of a method for recommending matching of scientific researchers and scientific research projects provided by the first embodiment of the present invention; Figure 2 is a flow chart of the steps of collecting and processing scientific researcher information in the first embodiment of the present invention; Figure 3 This is a flowchart of searching and querying papers and project experiences in the first embodiment of the present invention; Figure 4 This is a flowchart of summarizing and screening papers and project experiences in the first embodiment of the present invention; Figure 5 This is a flow chart of combining the screening of papers and project experience in the first embodiment of the present invention; Figure 6 is a flow chart of generating a public portrait in the first embodiment of the present invention; Figure 7 is a flow chart for generating an academic ability portrait in the first embodiment of the present invention; Figure 8 is a flow chart of a user preset character portrait in the first embodiment of the present invention; Figure 9 is a flow chart of a user generating a personalized portrait in the first embodiment of the present invention; Figure 10 is a flow chart for generating a project summary in a first embodiment of the present invention; Figure 11 This is a flowchart of basic demand matching in the first embodiment of the present invention; Figure 12 is a flowchart of academic needs matching in the first embodiment of the present invention; Figure 13 is a flow chart for generating project experience requirement matching in the first embodiment of the present invention; Figure 14 is a flowchart of matching report generation for researchers in the first embodiment of the present invention; Figure 15 This is a structural diagram of a matching recommendation system for scientific researchers and scientific research projects provided by the second embodiment of the present invention. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0022] The terms "first," "second," and the like in the specification and claims of this application are used to distinguish similar objects and are not used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate so that the embodiments of this application can be implemented in an order other than those illustrated or described herein. In addition, the term "and / or" in the specification and claims refers to at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0023] To facilitate the technical solutions of this application, some technical terms in the embodiments of this application are explained as follows.

[0024] Structured basic information of researchers: such as name, gender, professional title, degree, educational background, etc. Unstructured documents related to researchers: such as resumes, papers, project experience reports, etc. Semantic slicing: such as sliding window combined with natural paragraph boundary detection; Vectorized processing: such as using OpenAI's text-embedding-3-small model; The first type of agent: such as the paper summary agent and the project experience summary agent, which are based on large models such as GPT-4; Structured document summary data: such as a paper summary including research topics, methods, and results; a project experience summary including technical solutions and core technologies; The second type of agent: such as paper evaluation agent and project experience evaluation agent; Evaluation data: such as innovation, practicality, technical depth, and project completion; The third type of agent: project analysis agent, such as based on GPT-4o; Basic information of the research project: such as project level, title, expected results, cooperation mode, etc.; Project documents related to scientific research projects: such as application forms, reports, etc.; The fourth type of agent: the project requirements analysis agent, which can include three independent units (basic conditions, academic ability, and project experience) or one comprehensive unit; Structured project feature data: such as technical requirements, research objectives, research fields, keywords, core technologies, technical difficulties, innovations, etc. Basic requirements: such as job title, education background, age limit, etc., and preset rules for associated project levels; External academic data: such as journal name, impact factor, and author list; Preset rules: such as journal classification and impact factor; Author level: such as first author, corresponding author.

[0025] In order to illustrate the technical solution described in this application, specific embodiments are provided below.

[0026] Example 1 See also Figure 1 , which is a flow chart of a matching recommendation method for scientific researchers and scientific research projects proposed in the first embodiment of this application. The proposed method steps are as follows.

[0027] Step 1: Collect structured basic information of scientific researchers; receive and process unstructured documents related to scientific researchers, including semantic slicing and vectorization of document content, generating structured document summary data using the first type of intelligent agent, and evaluating the summary data using the second type of intelligent agent to generate evaluation data.

[0028] In this embodiment, the step of receiving and processing unstructured documents related to researchers further includes: Enhancement processing is performed on a specific type of unstructured document. The enhancement processing includes at least one or a combination of the following operations: Obtaining supplementary information related to a specific type of unstructured document from an external academic database or information source through a web crawler or API interface. The supplementary information includes at least one of the journal name, journal category, impact factor, author list, or project funding information in which the specific type of document is published; Determine the author level based on the author's signature information in a specific type of unstructured document input by the user or identified by the system; Based on supplementary information and / or author level, quality grading or tagging of specific types of unstructured documents is performed according to preset rules.

[0029] like Figure 2 As shown in Figure 1, step 1 is the process of collecting and processing scientific research personnel information. The specific process is as follows: 1. Create a Researcher: Users create a researcher profile under a specific unit (such as a college) through an interactive interface. The system provides a structured form, requiring users to fill in required information (name, gender, date of birth, and ID number) and optional information (degree, professional title, position, educational background, research focus, etc.). The system stores this information in a basic researcher information table in a MySQL relational database. Fields such as academic achievements and project experience are stored in JSON format to support dynamic expansion.

[0030] 2. Upload and process source files for the profile: Select a created researcher and upload relevant files (PDF, DOCX, etc.). When uploading, specify the file type (resume, paper, project history, or other).

[0031] File preprocessing: The system reads the file content. For paper files, the reference section can be selectively removed to optimize the subsequent summary.

[0032] Content slicing: Using a sliding window strategy (e.g., a window size of 1024 characters and a step size of 512 characters) combined with paragraph boundary detection, the file content is segmented into multiple semantically coherent text slices.

[0033] Vectorization and storage: Use a pre-trained vectorization model (such as OpenAI text-embedding-3-small) to convert each text slice into a 768-dimensional feature vector, and store the vector and its metadata (associated researcher ID, document ID, and slice sequence number) in a vector database.

[0034] 3. Generate paper summary and evaluation: After the user uploads the paper type file, the system prompts the user to supplement the paper's metadata: title, author type (first author, corresponding author, etc.), journal level (top journal, Class A, etc.).

[0035] Create a paper summary: Slice retrieval: Define a fixed query statement (such as "paper title, abstract, methods, results, conclusions") and vectorize it. Calculate the cosine similarity between the query vector and all slice vectors of the paper, and retrieve the top-K (for example, K = 10) most relevant slices.

[0036] Agent Summary: The retrieved Top-K slices are fed into the "Paper Summary Agent" (based on large models like GPT-4o, optimized with specific prompts). This agent generates a structured summary using a pre-set JSON template. The summary includes the following fields: all authors, journal name, key keywords, paper abstract, research topic, research question, research objectives and significance, core research methods, method logic, experimental results, and verification status.

[0037] Storage: The generated JSON format paper summary is stored in the paper summary table in the MySQL relational database, and the corresponding researchers and paper files are associated.

[0038] To create a paper evaluation: Input the newly generated paper summary JSON data into the "Paper Scoring Agent" and "Paper Extensibility Analysis Agent" (both can be implemented based on the big model).

[0039] Based on the summary content, the intelligent agent scores from preset dimensions and gives reasons (innovation, practicality, technicality, interdisciplinarity) and conducts an expansion analysis.

[0040] Generate evaluation data in JSON format and store it in the paper evaluation table in the MySQL relational database, and associate it with the corresponding paper summary.

[0041] 4. Generate project experience summary and evaluation: After the user uploads a project experience type file, the system prompts the user to supplement the metadata of the project experience: project name, project type, status, total budget, start and end dates, outcome type and description, project level (provincial key, national natural science, etc.), project role (person in charge, participant, etc.).

[0042] Creating a project experience summary: The process is similar to creating a paper summary. Use a fixed query (such as "project title, abstract, keywords, research background, technical solution, core technology, difficulties, and innovations") to retrieve relevant slices. Input these slices into the "Project Experience Summary Agent" to generate a structured JSON summary containing the relevant fields. This summary is stored in the MySQL project experience summary table.

[0043] Create a project experience review: The process is similar to creating a paper review. The generated project experience summary is fed into the "Project Experience Scoring Agent" and the "Project Experience Scalability Analysis Agent" for scoring based on pre-defined dimensions (project completion, technological innovation, goal achievement rate, and technological depth) and scalability analysis (technical risk, sustainability). The evaluation data is generated in JSON format and stored in the project experience evaluation table in MySQL.

[0044] Step 2: Based on structured basic information, processed unstructured documents, summary data and evaluation data, use the third type of intelligent agent to generate a portrait of the researcher. The portrait of the researcher includes a public portrait and / or an academic ability portrait for specific ability queries.

[0045] In this embodiment, generating an academic ability profile for a specific ability query specifically includes: Receive keywords input by users to describe specific academic abilities or research directions; Calling the keyword extraction agent to process user input, extract query keywords, and generate an expanded keyword set containing related terms; Using the query keyword and / or the extended keyword set as the retrieval basis, the screening and sorting module is called. The screening and sorting module uses a hybrid retrieval technology to search the structured document summary data of researchers, screen out relevant document summary data, and form a subset of relevant summary data; Calling the dimension generation agent to generate evaluation dimensions for assessing specific academic abilities based on the query intent input by the user; Call the academic ability portrait generation agent to generate an academic ability portrait report for specific query capabilities based on relevant summary data subsets and evaluation dimensions.

[0046] Furthermore, the screening and sorting module adopts hybrid retrieval technology, and its specific implementation includes at least: Perform full-text search: Use the database to index the pre-set text fields in the structured document summary data to find and recall document summary records that contain the query keywords; Furthermore, semantic retrieval is performed: query keywords or query statements are converted into query vectors, similarity is calculated between the query vectors and feature vectors of document summary data stored in the vector database, and semantically relevant document summary records are retrieved; The full-text search results and semantic search results are merged or sorted, and the relevant summary data subset is output.

[0047] Furthermore, this step also includes generating a brief description of the results, which is specifically implemented by: Categorize unstructured documents related to researchers into project and achievement information based on document content, metadata, or user-specified type tags; Structured storage of classified project and achievement information in the database, and execution of update and merge operations; When the need to generate a brief description of the results is triggered, the core field information of the project and results information under the name of the researcher is extracted from the database to form a structured information list; Call the large language model, take the structured information list as input, and generate a natural language text summary of the researcher's project experience and academic achievements based on preset prompts.

[0048] Step 2 is the process of generating a portrait of a researcher. The specific process is as follows: 1. Screening and sorting of paper / project experience summaries (e.g. Figures 3 to 5 As shown): This module is called when data needs to be prepared for generating portraits or matching projects.

[0049] Sorting: Summarizes all of a researcher's papers (or project experience) and sorts them according to pre-set rules. For example, papers might be sorted by: journal tier (top journal > Category A > ..., with different scores assigned) > author type (first author > corresponding author > ..., with different weights assigned) > total paper evaluation score (the sum of scores across all dimensions).

[0050] Filter (based on keywords): Full-text search: Create a full-text index in MySQL for the specified fields of the paper summary (core keywords, abstract, theme, etc.). Perform a fuzzy search using the user-provided keywords, and sort the returned results according to the above sorting rules, selecting the top-N results (e.g., N=5).

[0051] Semantic Search: Vectorize keywords. Retrieve three pre-generated vectors for all of the researcher's paper summaries from the vector database (core word vectors, core content vectors, and full-text summary vectors, generated by combining different fields). Calculate the weighted similarity between the keyword vector and each paper summary's three vectors (e.g., 0.5 core word similarity + 0.3 core content similarity + 0.2 full-text summary similarity). Sort by overall similarity, selecting the top-M vectors (e.g., M = 5).

[0052] Agent Screening: The combined results of the initial screening (full text + semantics) (without duplicates) are fed into the "Screening Agent" (Big Model). The Screening Agent determines the relevance of each summary based on the original keywords, removes those with low relevance, and generates a selection rationale for the retained summaries (e.g., "Paper A mentions BERT model optimization, with an 87% correlation to the target keyword 'Big Model'"). The final output is a list of paper summaries that meet the requirements and are limited in number (e.g., a maximum of five). The screening and sorting process for project experience summaries is similar.

[0053] 2. Generate character portraits (such as Figures 6 to 9 shown): (1) Public portraits Data source: basic information of researcher XX (from the database) + slice information of all his / her resume-type files.

[0054] Slice acquisition: After vectorizing a fixed query statement (such as "personal basic information, educational background, work experience, project experience, and skills expertise"), calculate the similarity with the resume file slice vector and retrieve highly relevant slices.

[0055] Agent generation: Input basic information and retrieved resume slices into the "public portrait generation agent" to generate a public portrait text containing personal profile, core skills, research direction, past experience, etc.

[0056] (2) Academic Ability Profile Trigger: The user asks the system through the interface, "What is the ability of researcher XX in [a certain aspect, such as large models]?" Keyword Extraction: The "Keyword Extraction Agent" parses the query, extracts core keywords, and generates extensions (LLM, generative AI, deep learning, etc.).

[0057] Data screening: Use the expanded keywords to call the "Paper / Project Experience Summary Screening and Sorting Module" to filter out the top-K paper summaries and top-L project experience summaries that are most relevant to the "big model".

[0058] Dimension generation: The "Dimension Generation Agent" dynamically generates evaluation dimensions and their explanations (such as technical capabilities, research influence, innovation and foresight) based on user queries (or keywords).

[0059] Portrait generation: The "academic ability portrait generation agent" receives the filtered summary data, dynamically generated dimensions and explanations, and generates a portrait description of the academic ability of researcher XX in this aspect (big model).

[0060] Brief description of projects and academic achievements hosted / participated in: Data update: When a user uploads project experience or paper files and completes the summary, the system automatically associates them with the "projects hosted / participated in" or "academic achievements" attributes of researcher XX (stored in the database and can be merged).

[0061] Brief description generation: When the "Brief Description of Achievements" function is called, the system reads XX's project and achievement list (core fields) from the database, and then calls the large model to convert the list information into a brief description text described in natural language.

[0062] Step 3: Collect basic information of scientific research projects; receive and process project documents related to scientific research projects, use the fourth type of intelligent agent to analyze the content of project documents and basic project information, generate structured project feature data and hierarchical project requirements, and project requirements include at least basic requirements, academic requirements, and project experience requirements.

[0063] like Figure 10 As shown in Figure 3, step 3 is the process of collecting and analyzing scientific research project information. The specific process is as follows: 1. Create a research project: Users create a project YY through the interface. Fill in the required information (project ID, title, expected outcomes, collaboration method, planned start and end dates, status, etc.). The system automatically associates some basic rules based on the project level. Some fields (such as technical requirements and research objectives) are initially blank, pending subsequent AI analysis and generation.

[0064] 2. Upload and process project source files: Users upload relevant files for Project YY (application, final report, etc.). The system slices and vectorizes them, storing them in the vector database. The process is the same as document processing in step 1.

[0065] 3. Project field analysis: Slice retrieval: After vectorization, a fixed query statement (such as "extract key information such as project summary, technical requirements, research objectives, core technologies, difficulties, and innovations") is used to calculate the similarity with the slice vector of the project source file and retrieve highly relevant slices.

[0066] Agent Analysis: The retrieved slices are fed into the Project Analysis Agent (based on GPT-4o). Based on pre-set templates or instructions, the agent analyzes and extracts information, generating JSON-formatted data containing: technical requirements, research objectives, research areas, keywords, core technologies, technical difficulties, innovations, and required tools / platforms.

[0067] Update storage: Update the generated JSON data to the corresponding field of project YY in the MySQL database.

[0068] 4. Generate project requirements analysis: Re-retrieve slices (optional, or reuse the results of the previous step): Use a query statement that focuses more on the requirements to retrieve related slices.

[0069] Agent summary and analysis: The retrieved slices and basic information of project YY in the database and the newly generated field information are input into the "Project Requirements Analysis Agent" (or its three sub-agents: basic requirements analysis, academic requirements analysis, and project experience requirements analysis).

[0070] Generate requirements: The agent generates three structured requirement descriptions: Project summary (optional, for overall understanding) Academic needs analysis (describe the theoretical knowledge, research direction, technical capabilities, etc. required for the project) Project experience requirements analysis (describe the similar project experience required for the project, specific technical route practical experience, achievement requirements, etc.) Storage: Store the generated demand analysis results in the MySQL database and associate them with project YY.

[0071] Step 4: Receive the research project to be matched and at least one research researcher to be matched; perform the single-person matching stage for each research researcher to be matched, and the matching stage includes at least basic rule matching based on the basic requirements of the research project to be matched and the basic information of the research researcher to be matched, academic requirement matching based on the academic requirements of the project and the research researcher portrait and related document summary, and project experience requirement matching based on the project experience requirements of the research project to be matched and the research researcher portrait; perform the multi-person comparison and ranking stage for the research researchers who pass the single-person matching stage to generate recommended ranking results.

[0072] In this embodiment, in the single-person matching stage of this step, before performing academic requirement matching and project experience requirement matching, the step further includes performing an information preparation sub-step, which at least includes: Use the project feature data or project demand analysis results of the scientific research project to be matched as query input; Search and sort the personal knowledge base of the current researcher to be matched, and filter out a subset of document summary data related to the research project to be matched; Determine whether the scientific researcher to be matched already has an academic ability profile related to the abilities required by the matching project. If not, generate an academic ability profile for preparation based on the keywords of the matching project; The filtered subset of relevant summaries and the prepared relevant profiles are used as input to the subsequent matching step.

[0073] In this embodiment, the multi-person comparison and ranking stage of this step specifically includes: Perform the global evaluation sub-step: For each candidate researcher, extract the key academic achievements and project experience indicators preset in the candidate knowledge base; calculate the global ability evaluation score based on the extracted indicators according to the preset evaluation model; Execute the comprehensive ranking sub-step: comprehensively consider the academic needs matching score, project experience needs matching score and calculated global ability evaluation score obtained by each candidate researcher in the single matching stage, calculate the comprehensive matching total score through the preset algorithm; and rank all candidates according to the comprehensive matching total score.

[0074] like Figures 11 to 14 As shown, step 3 is the matching recommendation process, the specific process is as follows: 1. Initiate matching: The user selects a project YY to be matched and selects one or more (such as XX, XY, XZ) researchers to be matched.

[0075] 2. Single person matching (performed separately for XX, XY, and XZ): Take the matching of researcher XX to project YY as an example: Get project information: Basic screening rules: obtain admission rules (must be met), exclusion rules (must not be met), and verification rules (need to be verified offline) from the project level associated with project YY.

[0076] Project Knowledge Base: Obtain basic information about Project YY, AI-generated fields (keywords, technical requirements, etc.), and the latest academic needs analysis and project experience needs analysis results.

[0077] Get information about researchers: Basic Information: Obtain basic information of XX (used for basic rule matching).

[0078] Relevant summary screening: Use the keywords of project YY to call the "Paper / Project Experience Summary Screening and Sorting Module" to filter out the most relevant paper summaries and project experience summaries (such as the top 5) from XX's knowledge base.

[0079] Prepare relevant profiles: Invoke the "Profile Identification Agent" to determine whether XX's existing, generated profiles contain profiles related to the skills required for Project YY. If not, temporarily invoke the "Character Profile Generation Module" based on Project YY's keywords to generate an academic profile for XX related to Project YY.

[0080] Perform a match: Phase 1: Basic Rule Matching. Using a "Basic Screening Rule Matching Agent" (or a rule engine), enter XX's basic information and Project YY's basic screening rules. If the entry rules are not met or the exclusion rules are met, XX is deemed ineligible for participation and the process terminates. Otherwise, the process proceeds to the next phase.

[0081] Phase 2: Academic Needs Matching. Using the "Academic Needs Matching Agent," input is XX's basic information, relevant (existing or improvised) character profiles, selected relevant paper summaries, and Project YY's basic information and academic needs analysis results. The agent determines XX's academic compatibility with the project and outputs a score and justification.

[0082] Phase 3: Project Experience and Requirements Matching. Using the "Project Experience and Requirements Matching Agent," input XX's basic information, related profiles, selected project experience summaries, and Project YY's basic information and project experience and requirements analysis results. The agent determines the match between XX's project experience and the project, and outputs a score and justification.

[0083] Individual Result Determination: XX is considered preliminarily eligible for the program only if he / she passes the basic matching criteria and meets certain thresholds for both academic and program experience requirements (or is determined to be suitable). Record the matching results, scores, and rationale for each stage.

[0084] 3. Multi-person comparison and sorting: For all researchers who have passed the individual matching stage (assuming XX and XY passed, but XZ failed): Global evaluation: For XX and XY, the system extracts all or key achievements in their knowledge base (such as the number of top journal papers, the number of national-level projects they have presided over, etc.) and calculates a comprehensive ability score based on preset weights.

[0085] Comprehensive ranking: XX and XY's academic and project experience scores from the individual matching phase are combined to calculate their combined scores and rank them. If their combined scores are similar, a secondary ranking may be performed based on objective indicators (e.g., XX has a top-tier journal publication, while XY does not).

[0086] Generate a report: The system generates a detailed match report for XX and XY, including basic rule pass status, academic requirement match details (score, justification), project experience match details (score, justification), global evaluation metrics, and the final recommended ranking. This report is presented to the user to assist in decision-making. Based on the report, combined with verification rules (offline verification) and actual circumstances, the user can finalize the project candidate.

[0087] Example 2 See also Figure 15 , shown is a schematic diagram of the structure of a matching recommendation system for scientific researchers and scientific research projects proposed in the second embodiment of this application, the system comprising: The researcher information processing module 100 is used to collect structured basic information of researchers; receive and process unstructured documents related to the researchers, said processing including semantic slicing and vectorizing the document content, generating structured document summary data using the first type of intelligent agent, and evaluating the summary data using the second type of intelligent agent to generate evaluation data; A researcher portrait generation module 200 is configured to generate a researcher portrait using a third type of agent based on the structured basic information, the processed unstructured documents, the summary data, and the evaluation data. The researcher portrait includes a public portrait and / or an academic ability portrait for a specific ability query. The scientific research project information processing module 300 is used to collect basic information about scientific research projects; receive and process project documents related to the scientific research projects; use the fourth type of agent to analyze the content of the project documents and the basic project information, and generate structured project feature data and hierarchical project requirements, which include at least basic requirements, academic requirements, and project experience requirements; The matching recommendation module 400 is used to receive a scientific research project to be matched and at least one scientific researcher to be matched; a single matching stage is performed for each scientific researcher to be matched, and the matching stage includes at least basic rule matching based on the basic requirements of the scientific research project to be matched and the basic information of the scientific researcher to be matched, academic requirement matching based on the academic requirements of the project and the scientific researcher portrait and related document summary, and project experience requirement matching based on the project experience requirements of the scientific research project to be matched and the scientific researcher portrait; a multi-person comparison and ranking stage is performed for the scientific researchers who pass the single matching stage to generate a recommended ranking result.

[0088] In the embodiments of the present application, a matching recommendation system for scientific researchers and research projects can be a device, or a component, integrated circuit, or chip in a terminal. The device can be a mobile electronic device or a non-mobile electronic device. For example, the mobile electronic device can be a mobile phone, tablet computer, laptop computer, PDA, in-vehicle electronic device, wearable device, ultra-mobile personal computer (UMPC), netbook, or personal digital assistant (PDA), etc. The non-mobile electronic device can be a server, network attached storage (NAS), personal computer (PC), etc., which are not specifically limited in the embodiments of the present application.

[0089] In an embodiment of the present application, a matching recommendation system for scientific researchers and scientific research projects may be a device having an operating system. The operating system may be an Android operating system, an iOS operating system, or other possible operating systems, which are not specifically limited in the embodiment of the present application.

[0090] The matching recommendation system for scientific researchers and scientific research projects provided in the embodiment of the present application can achieve Figure 1 In the method embodiment, each process of implementing a method for matching and recommending scientific researchers and scientific research projects will not be described here in detail to avoid repetition.

[0091] Optionally, an embodiment of the present application also provides an electronic device, including a processor, a memory, and a program or instruction stored in the memory and executable on the processor. When the program or instruction is executed by the processor, each process of the above-mentioned embodiment of the method for recommending matching scientific researchers and scientific research projects is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0092] An embodiment of the present application also provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, each process of the above-mentioned embodiment of the matching recommendation method for scientific researchers and scientific research projects is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0093] The processor is the processor in the electronic device described in the above embodiment. The readable storage medium includes a computer-readable storage medium, such as a computer read-only memory (ROM), random access memory (RAM), a magnetic disk, or an optical disk.

[0094] It should be noted that, in this article, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element. In addition, it should be noted that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in the opposite order according to the functions involved. For example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0095] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a more preferred embodiment. Based on this understanding, the technical solution of this application, or the part that contributes to the existing technology, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of this application.

[0096] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of this application, ordinary technicians in this field can also make many forms without departing from the purpose of this application and the scope of protection of the claims, all of which are within the protection of this application.

Claims

1. A matching recommendation method for scientific researchers and scientific research projects, characterized in that: The matching recommendation method includes: Researcher information collection and processing steps: collecting structured basic information of researchers; receiving and processing unstructured documents related to the researchers, said processing including semantic slicing and vectorizing the document content, generating structured document summary data using the first type of intelligent agent, and evaluating the summary data using the second type of intelligent agent to generate evaluation data; A researcher portrait generation step: based on the structured basic information, the processed unstructured documents, the summary data, and the evaluation data, a third-type intelligent agent is used to generate a researcher portrait, wherein the researcher portrait includes a public portrait and / or an academic ability portrait for a specific ability query; Research project information collection and analysis steps: Collect basic information about the research project; receive and process project documents related to the research project, use the fourth type of agent to analyze the content of the project documents and the basic project information, and generate structured project feature data and hierarchical project requirements, which include at least basic requirements, academic requirements, and project experience requirements; Matching recommendation steps: receive the scientific research project to be matched and at least one scientific researcher to be matched; perform the single matching stage for each scientific researcher to be matched, the matching stage at least includes basic rule matching based on the basic requirements of the scientific research project to be matched and the basic information of the scientific researcher to be matched, academic requirements matching based on the academic requirements of the project and the scientific researcher portrait and related document summary, and project experience requirement matching based on the project experience requirements of the scientific research project to be matched and the scientific researcher portrait; perform the multi-person comparison and ranking stage for the scientific researchers who pass the single matching stage to generate recommended ranking results.

2. A matching recommendation method for scientific researchers and scientific research projects according to claim 1, characterized in that: In the step of collecting and processing the information of the scientific researchers, the step of receiving and processing the unstructured documents related to the scientific researchers further includes: Enhancement processing is performed on a specific type of unstructured document, where the enhancement processing includes at least one or a combination of the following operations: Obtaining supplementary information related to a specific type of unstructured document from an external academic database or information source through a web crawler or API interface, wherein the supplementary information includes at least one of the journal name, journal category, impact factor, author list, or project funding information in which the specific type of document is published; Determine the author level based on the author's signature information in a specific type of unstructured document input by the user or identified by the system; Based on the supplementary information and / or the author level, and in accordance with preset rules, a quality grade or tag is performed on a specific type of unstructured document.

3. A matching recommendation method for scientific researchers and scientific research projects according to claim 1 or 2, characterized in that: In the step of generating a scientific researcher portrait, generating the academic ability portrait for the specific ability query specifically includes: Receive keywords input by users to describe specific academic abilities or research directions; Invoking the keyword extraction agent to process the user input, extract query keywords, and generate an expanded keyword set containing related terms; Using the query keyword and / or the extended keyword set as a retrieval basis, calling a screening and sorting module, the screening and sorting module using a hybrid retrieval technology to search the structured document summary data of the researcher, screening out relevant document summary data, and forming a relevant summary data subset; Invoking a dimension generation agent to generate an evaluation dimension for evaluating the specific academic ability according to the query intention input by the user; Call the academic ability portrait generation agent to generate an academic ability portrait report for the specific query ability based on the relevant summary data subset and the evaluation dimension.

4. The method for recommending matching of scientific researchers and scientific research projects according to claim 3, characterized in that: The screening and sorting module adopts a hybrid retrieval technology, and its specific implementation includes at least: Perform full-text search: use the database to create an index on the preset text fields in the structured document summary data to find and recall the document summary records containing the query keywords; Furthermore, semantic retrieval is performed: the query keyword or query statement is converted into a query vector, similarity is calculated between the query vector and the feature vector of the document summary data stored in the vector database, and semantically relevant document summary records are retrieved; The full-text search results and semantic search results are merged or sorted, and the relevant summary data subset is output.

5. A matching recommendation method for scientific researchers and scientific research projects according to any one of claims 1 to 4, characterized in that: The step of generating a scientific researcher portrait also includes generating a brief description of the results, which is specifically implemented as follows: Categorize unstructured documents related to researchers into project and achievement information based on document content, metadata, or user-specified type tags; Structured storage of classified project and achievement information in the database, and execution of update and merge operations; When the need to generate a brief description of the results is triggered, the core field information of the project and results information under the name of the researcher is extracted from the database to form a structured information list; The large language model is called, the structured information list is used as input, and according to preset prompts, a natural language text summary of the project experience and academic achievements of the researcher is generated.

6. A method for recommending matching of scientific researchers and scientific research projects according to any one of claims 1 to 5, characterized in that: In the single-person matching stage of the matching recommendation step, before performing the academic needs matching and the project experience needs matching, the step further includes performing an information preparation sub-step, which at least includes: Using the project feature data or project demand analysis results of the scientific research project to be matched as query input; Search and sort the personal knowledge base of the current researcher to be matched, and filter out a subset of document summary data related to the research project to be matched; Determine whether the scientific researcher to be matched already has an academic ability profile related to the ability required by the project to be matched. If not, generate an academic ability profile for preparation based on the keywords of the project to be matched; The filtered subset of relevant summaries and the prepared relevant profiles are used as input to the subsequent matching step.

7. A method for recommending matching of scientific researchers and scientific research projects according to any one of claims 1 to 6, characterized in that: The multi-person comparison and ranking stage of the match recommendation step specifically includes: Perform the global evaluation sub-step: For each candidate researcher, extract the key academic achievements and project experience indicators preset in the candidate knowledge base; calculate the global ability evaluation score based on the extracted indicators according to the preset evaluation model; Execute the comprehensive ranking sub-step: comprehensively consider the academic needs matching score, project experience needs matching score and calculated global ability evaluation score obtained by each candidate researcher in the single matching stage, calculate the comprehensive matching total score through the preset algorithm; and rank all candidates according to the comprehensive matching total score.

8. A matching recommendation system for scientific researchers and scientific research projects, characterized in that: The matching recommendation system includes: A researcher information processing module is used to collect structured basic information about researchers; receive and process unstructured documents related to the researchers, including semantic slicing and vectorizing the document content, generating structured document summary data using the first type of intelligent agent, and evaluating the summary data using the second type of intelligent agent to generate evaluation data; A researcher portrait generation module is configured to generate a researcher portrait using a third type of agent based on the structured basic information, the processed unstructured documents, the summary data, and the evaluation data. The researcher portrait includes a public portrait and / or an academic ability portrait for a specific ability query. The scientific research project information processing module is used to collect basic information about scientific research projects; receive and process project documents related to the scientific research projects, analyze the content of the project documents and basic project information using the fourth type of agent, and generate structured project feature data and hierarchical project requirements, which include at least basic requirements, academic requirements, and project experience requirements; The matching recommendation module is used to receive scientific research projects to be matched and at least one scientific researcher to be matched; a single matching stage is performed for each scientific researcher to be matched, and the matching stage at least includes basic rule matching based on the basic requirements of the scientific research project to be matched and the basic information of the scientific researcher to be matched, academic requirements matching based on the academic requirements of the project and the scientific researcher portrait and related document summary, and project experience requirement matching based on the project experience requirements of the scientific research project to be matched and the scientific researcher portrait; a multi-person comparison and ranking stage is performed for the scientific researchers who pass the single matching stage to generate a recommended ranking result.

9. An electronic device, characterized in that: It includes a processor, a memory, and a program or instruction stored in the memory and executable on the processor. When the program or instruction is executed by the processor, the steps of a method for recommending matching scientific researchers and scientific research projects as described in any one of claims 1 to 7 are implemented.

10. A readable storage medium, characterized in that: The readable storage medium stores a program or instruction, and when the program or instruction is executed by the processor, the steps of the method for matching scientific researchers and scientific research projects as described in any one of claims 1 to 7 are implemented.

Images