A scientific and technological achievement transformation service system and a data processing method thereof

Abstract

This invention relates to the field of data processing systems or methods for the transformation of scientific and technological achievements. Specifically, it provides a service system for the transformation of scientific and technological achievements and its data processing method. Through multi-source data fusion, dynamic risk assessment and intelligent matching, it achieves full-chain coverage, dynamic optimization and visualization of the transformation of scientific and technological achievements. From data collection to commercialization, it seamlessly connects each link and realizes full-process management from technology evaluation to commercialization. Through TRL model, knowledge graph and blockchain technology, it improves the efficiency and success rate of transformation. It is suitable for technology transfer scenarios in universities, research institutions and enterprises.

Description

Technical Field

[0001] This invention relates to the field of data processing systems or methods for the transformation of scientific and technological achievements, and specifically provides a service system for the transformation of scientific and technological achievements and its data processing method. Background Technology

[0002] In recent years, various regions have actively explored ways to enhance the transformation of cutting-edge major scientific and technological achievements, accelerate the industrialization of major scientific and technological achievements, and empower the development and growth of new productive forces through scientific and technological innovation.

[0003] Most existing information operation platforms for scientific and technological achievements (such as patent database search tools commonly used in the industry) mainly function as partial-dimensional search tools, lacking intelligent evaluation and dynamic planning capabilities. This leads to the following problems in the current transformation of achievements: severe data silos: scientific research data (patents, papers), market data, and enterprise needs are scattered across different platforms, lacking unified analysis tools; strong subjectivity in evaluation: traditional reliance on expert experience is inefficient and makes it difficult to quantify the technological value and risks; inefficient resource matching: information asymmetry between technology suppliers and demanders results in long transformation cycles and low success rates.

[0004] Therefore, there is an urgent need for a systematic solution that integrates data fusion, intelligent analysis, and full-process management. Summary of the Invention

[0005] This invention provides a technology transfer service system to address the shortcomings of existing technologies, such as severe data silos, subjective evaluation, and inefficient resource matching, which lack intelligent evaluation and dynamic planning capabilities. It provides a systematic solution for data fusion, intelligent analysis, and full-process management.

[0006] In a first aspect, the present invention provides a technology transfer service system, comprising: a multi-source data processing module, an analysis and evaluation module, a resource matching module, a visualization module, and a full-process management and control module, to achieve full-process information support from data collection, analysis, and evaluation of scientific and technological achievements to final commercial application; The multi-source data processing module uses a data acquisition component to collect multi-source data using acquisition tools, and uses a data standardization component to standardize and clean the collected multi-source data, which is used to integrate scientific research data, market data and resource data of scientific and technological achievements. The analysis and evaluation module constructs an evaluation model based on knowledge graph (Neo4j) association technology and industry demand nodes to generate a visualized technology-market mapping relationship, which is used to quantify the technology maturity (TRL), market fit and risk of scientific and technological achievements. The resource matching module uses collaborative filtering algorithm to calculate enterprise matching degree. It defines matching logic through rule engine (Drools) to intelligently recommend cooperative enterprises, investors and transformation paths of scientific and technological achievements. The visualization module includes a front-end and a back-end. The front-end is used to generate interactive charts (such as technology maturity radar charts and market growth curves), while the back-end integrates Monte Carlo simulation to predict the economic benefits of different paths, and is used to dynamically display the evaluation results of scientific and technological achievements, market trends, and the revenue and risk of simulated transformation paths. The full-process management module manages multi-stage task dependencies through a workflow engine, ensures data immutability through blockchain-based evidence storage of key nodes, tracks the progress of technology transfer, and automatically generates compliance documents for technology achievements. The aforementioned scientific and technological achievements are practically valuable results generated through scientific research and technological development. They are knowledge products with recognized academic or economic value, obtained by individuals, research institutions, universities, or other innovation entities through complex intellectual labor in scientific and technological activities. These include, but are not limited to, technical solutions, materials, and their intellectual property rights that have not been implemented, produced, or industrialized on a large scale, such as new inventions, innovative products, technical methods, processes, experimental data, technical standards, patents, papers, and software programs. The technology transfer process involves matching and recommending partner companies, investors, and transfer pathways for the aforementioned scientific and technological achievements, and completing their large-scale implementation, production, and industrialization. The innovation entities include research institutions, research institutes, universities, or other organizations or individuals that possess the aforementioned scientific and technological achievements. The input parameters of the evaluation model include patent citation count, experimental success rate, and market demand growth rate. The technological achievements are fused through multimodal data from the multi-source data processing module, mapping technical data, market demand, and resource networks into a unified semantic space. The analysis and evaluation module's assessment model, combined with technology maturity levels, patent portfolios, and market competitors, generates a risk score in real time. The resource matching module's intelligent matching engine, driven by both collaborative filtering and rule-based engines, intelligently recommends partner companies, investors, and transformation paths, improving the accuracy of resource matching. The resources refer to potential partner companies, investors, or incubators of the technological achievements.

[0007] In embodiments of the present invention, the data acquisition component connects to the internal database of the innovation entity and the enterprise demand platform through an API (Application Programming Interface). The natural language processing component cleans unstructured data (such as technical documents and semantic parsing of patent abstracts), integrates scientific research data (such as patents, papers, and experimental data), market data (such as industry demand, competitor analysis, industry reports, and enterprise demand), intellectual property data (such as patent status and legal risks), and resource data (such as cooperative resource data from enterprises, investors, etc., hereinafter the same), and performs data standardization processing (such as unifying the format, deduplication, and completion). The analysis and evaluation module uses evaluation models (such as those based on random forests / deep learning) to conduct technology maturity assessment, market fit analysis, and risk assessment. The technology maturity assessment is based on the TRL model (Technology Maturity Model) to quantify the feasibility of scientific and technological achievements. The market fit analysis uses machine learning to predict the commercialization potential of the technology (such as market size and competitive barriers). The risk assessment is used to identify intellectual property risks (such as the possibility of infringement) and technological bottlenecks (such as the difficulty of mass production). The resource matching module recommends matching resources through a collaborative filtering algorithm and defines matching logic (such as matching enterprise technology needs keywords with patent abstracts and keyword weight allocation) through a rule engine to intelligently match scientific and technological achievements with potential cooperative enterprises, investors or incubators, and generate customized transformation path suggestions (such as licensing, joint development, and startup incubation). The visualization module provides an interactive dashboard on the backend, displaying technology evaluation results, market trends and resource matching degree, and supports dynamic simulation of the benefits and risks of different conversion paths (such as Monte Carlo simulation). The full-process management module is used to track the progress of technology transfer (such as cooperation negotiation, contract signing, pilot testing, and mass production), and automatically generate technology transfer reports and compliance documents (such as technology licensing contract templates).

[0008] In embodiments of the present invention, the data acquisition component employs a distributed crawler framework to crawl public databases and obtain unstructured data; the data standardization component uses a natural language processing (NLP) component to perform data standardization and cleaning on the acquired unstructured data; the multi-source data includes: scientific research data, market data, and resource data; the scientific research data includes: patent databases, paper repositories, and laboratory management systems; the market data includes: industry reports and enterprise demand data from enterprise demand platforms; the resource data includes: business directories and investor directories; the acquisition tools include: web crawlers and API integration; the API integration calls third-party APIs through the Requests library; The data standardization and cleaning includes: data format unification and data cleaning; the data format unification includes: converting patent data into JSON format (JavaScript Object Notation, an open standard file format and data exchange format); converting experimental data into CSV format (Comma-Separated Values, a common file format widely used in business and science); the data cleaning includes: deduplication and missing value completion; the deduplication is based on unique identifiers (such as patent numbers) using Pandas (an open-source data analysis library based on Python); the missing value completion uses KNN (K-Nearest Neighbors) interpolation.

[0009] In embodiments of the present invention, the analysis and evaluation module is used to perform technology maturity assessment, market fit analysis, and knowledge graph construction; the technology maturity assessment adopts a TRL model to determine the technology maturity level based on the experimental stage; the TRL model includes a rule engine, which automatically determines the level based on experimental data; the market fit analysis adopts a machine learning model, and the application of the machine learning model includes: input features and model training; The input features include: patent citation count, TF-IDF value of technical keywords (a value calculated by the TF-IDF algorithm to measure the importance of a word in a document; TF-IDF (Term Frequency-Inverse Document Frequency) is a statistical method that evaluates the importance of a word to a document by combining term frequency (TF) and inverse document frequency (IDF), and market demand growth rate); the model training uses random forest to predict commercial potential; the knowledge graph construction uses knowledge graph database tools to define nodes and relationships.

[0010] In embodiments of the present invention, the resource matching module includes a collaborative filtering algorithm; the implementation steps of the collaborative filtering algorithm include: constructing an enterprise-technology matrix, the matrix value being a matching score; using the Surprise library (a Python recommendation system library specifically designed for building and analyzing recommendation systems with explicit scoring data) to implement collaborative filtering; the end-to-end management module includes blockchain notarization; the blockchain notarization is used for smart contracts.

[0011] The technology transfer service system of this invention, through multi-source data fusion, dynamic risk assessment and intelligent matching engine, can achieve full-chain coverage, dynamic optimization and transparency and trustworthiness, realizing seamless connection of each link from data collection to commercialization, realizing full-process management from technology evaluation to commercialization, combining TRL model, knowledge graph and blockchain technology to improve conversion efficiency and success rate, continuously improving the accuracy of evaluation and matching through machine learning, and blockchain notarization to ensure traceability and tamper-proof of the conversion process, significantly improving the efficiency and success rate of technology transfer, and is suitable for technology transfer scenarios in universities, research institutions and enterprises.

[0012] Secondly, the present invention also provides a method for processing data related to the transformation of scientific and technological achievements, applicable to any of the aforementioned scientific and technological achievement transformation service systems, comprising: Acquire scientific research data, market data, intellectual property data, and collaborative resource data, and perform data fusion and standardization processing; By constructing an evaluation model, we can conduct technology maturity assessment, market fit analysis, and risk assessment on the data after fusion and standardization. Based on the evaluation and analysis results, the system intelligently matches scientific and technological achievements with potential partner companies, investors, or incubators through collaborative filtering algorithms and rule engines, generating customized transformation path suggestions.

[0013] In embodiments of the present invention, the acquisition of scientific research data, market data, intellectual property data, and collaborative resource data, and the subsequent data fusion and standardization processing, includes: connecting to the internal database of the innovation entity and the enterprise demand platform via API; cleaning unstructured data (such as technical documents and semantic parsing of patent abstracts) using natural language processing components; integrating scientific research data (such as patents, papers, and experimental data), market data (such as industry demand, competitor analysis, industry reports, and enterprise demand), intellectual property data (such as patent status and legal risks), and collaborative resource data (such as enterprises and investors); and performing data standardization processing (such as unifying formats, deduplication, and completion). The process of conducting technology maturity assessment, market fit analysis, and risk assessment on the fused and standardized data by constructing an evaluation model includes: conducting technology maturity assessment, market fit analysis, and risk assessment by constructing an evaluation model (such as based on random forest / deep learning); the technology maturity assessment is based on the TRL model to quantify the feasibility of scientific and technological achievements; the market fit analysis uses machine learning to predict the commercialization potential of the technology (such as market size and competitive barriers); and the risk assessment is used to identify intellectual property risks (such as the possibility of infringement) and technological bottlenecks (such as the difficulty of mass production). Based on the evaluation and analysis results, the system intelligently matches scientific and technological achievements with potential partner companies, investors, or incubators through collaborative filtering algorithms and rule engines, generating customized transformation path suggestions. This includes: recommending matching resources through collaborative filtering algorithms; defining matching logic through rule engines (such as matching enterprise technology needs keywords with patent abstracts and keyword weight allocation); intelligently matching scientific and technological achievements with potential partner companies, investors, or incubators; and generating customized transformation path suggestions (such as licensing, joint development, and startup incubation).

[0014] In embodiments of the present invention, scientific research data, market data, intellectual property data, and collaborative resource data are acquired, and data fusion and standardization processing is performed, including: mapping data from different sources to the same semantic space, involving NLP technology and feature engineering; and vectorizing and aligning keywords in patent texts and industry reports. The inputs to the evaluation model include TRL, patent coverage, and competitor technology layout, and the output includes a risk score (e.g., 0-100 points), which mainly involves the impact of different weights, as well as the model training process, such as the algorithm used, data sources, and verification methods. The collaborative filtering includes: constructing a user-technology matrix to handle cold starts; the rule engine includes keyword matching logic to improve matching accuracy; the intelligent matching of scientific and technological achievements with potential partners, investors or incubators includes: calculating the enterprise-technology matching degree.

[0015] In embodiments of the present invention, the acquisition of scientific research data, market data, intellectual property data, and cooperative resource data, and the data fusion and standardization processing include: cross-modal data alignment and active learning annotation; the cross-modal data alignment is used to map technical parameters (such as material properties) and market demands (such as industry pain points) to a unified semantic space; the active learning annotation optimizes the evaluation model through expert feedback (such as annotating high-potential technology cases).

[0016] In embodiments of the present invention, scientific research data, market data, intellectual property data, and cooperative resource data are acquired and subjected to data fusion and standardization processing, including: integrating multi-source heterogeneous data (such as technical parameters, market demand, network relationships of cooperative resources, etc.) through cross-modal data alignment and active learning annotation to improve data consistency and information density; The cross-modal data alignment includes: semantic embedding, knowledge graph alignment, and loss function design; the semantic embedding includes: using a pre-trained model to convert text data (such as patent abstracts and industry reports) into vectors; and standardizing structured data (such as experimental parameters and market size values) (Z-score) and concatenating them into the same vector space; The knowledge graph alignment includes: defining cross-modal node relationships in the knowledge graph; The loss function design includes: using contrastive learning to minimize the distance between similar data and maximize the distance between dissimilar data; The process of active learning and labeling includes: the initial model is trained using a small amount of labeled data (such as 100 technical-market matching samples); the model predicts on unlabeled data and selects the sample with the highest uncertainty (such as a prediction probability close to 0.5); after expert labeling, the sample is added to the training set, and the model is iteratively optimized.

[0017] In embodiments of the present invention, the evaluation model quantifies technology transformation risk in real time through input parameters and their calculation, risk scoring, and model training, and supports dynamic adjustment of weights and thresholds. The input parameters and their calculation include: technology maturity, patent coverage, and competitor technology layout. The technology maturity is determined based on the experimental stage (e.g., level 1-9). The patent coverage is calculated by the number of countries covered by the patent family (e.g., PCT application covering 50 countries → coverage = 50 / 195 ≈ 25.6%). The competitor technology layout includes: calculating technology similarity (e.g., cosine similarity) through patent citation networks. The model training includes: training data and model selection; the training data comes from historical conversion project data (e.g., features such as TRL, patent coverage, number of competitors; labels such as success / failure); the model selection is to choose a gradient boosting tree (XGBoost), which supports feature importance analysis; The intelligent matching accurately matches scientific and technological achievements with enterprise needs through collaborative filtering (user-technology matrix), the application of the rule engine (Drools), and hybrid matching, supporting the fusion of multiple strategies. The collaborative filtering includes matrix construction and matrix completion. The application of the rule engine mainly includes the definition and execution of the rule engine (Drools). The matrix completion uses SVD decomposition (Singular Value Decomposition, an important linear algebra matrix decomposition term, which is a generalization of eigenvalue decomposition on any matrix) to predict missing values. The execution process of the rule engine includes: inputting technical features and enterprise features into the rule engine; triggering all matching rules and generating a candidate matching list; sorting by score and outputting Top-N recommendation results (recommendation results obtained by Top-N analysis method, which obtains the required N data points from the research object through the Top-N algorithm and selects the largest or smallest N data points from the sorted list). The hybrid matching strategy includes: cold start processing; the cold start processing includes: when a new enterprise / technology has no historical data, relying only on the rule engine and keyword matching; Thirdly, the present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement any of the aforementioned technology transfer service systems.

[0018] The beneficial effects of this invention are as follows: By integrating and standardizing scientific research data, market data, intellectual property data, and collaborative resource data, and through risk assessment and intelligent matching, this invention utilizes cross-modal semantic alignment, risk quantification models, and hybrid matching strategies to achieve a closed loop from data to decision-making and full-process management from technology evaluation to commercialization. It innovatively combines TRL models, knowledge graphs, and blockchain technology, significantly improving conversion efficiency and success rate, and is suitable for technology transfer scenarios in universities, research institutions, and enterprises.

[0019] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0021] Figure 1 This is one of the block diagrams of the technology transfer service system provided in the embodiments of the present invention; Figure 2 This is the second block diagram of the technology transfer service system provided in this embodiment of the invention; Figure 3 This is the third block diagram of the technology transfer service system provided in this embodiment of the invention; Figure 4 This is a flowchart of the data processing method for the transformation of research results provided in this embodiment of the invention; Figure 5 This is a flowchart of data fusion and standardization processing provided in an embodiment of the present invention; Figure 6 This is an application flowchart of the evaluation model provided in the embodiments of the present invention; Figure 7 This is a schematic diagram of the intelligent matching process provided in an embodiment of the present invention; Figure 8 This is an execution flowchart of the rule engine provided in an embodiment of the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0023] Technology transfer typically involves transforming scientific and technological achievements (such as patents, technologies, and inventions) into practical applications or commercial products. The multi-source data involved in this process includes technical parameters of the scientific and technological achievements, market demand analysis, intellectual property information, and collaborative resources. Therefore, the technology transfer service system in this embodiment needs to possess functions such as data integration, intelligent analysis, risk assessment, and matching. In the specific implementation of this solution, multi-source data collection utilizes web crawling and API integration; the intelligent analysis module employs machine learning and knowledge graphs, involving model selection and training methods; resource matching uses collaborative filtering or rule engines; visualization utilizes a front-end framework and data visualization library; and end-to-end management employs a workflow engine and blockchain. This requires implementation via computers and the internet, using open-source or publicly available functional library programs.

[0024] The following is combined Figures 1 to 8 The embodiments shown describe the technical solution of the present invention in detail: Example 1: This example provides a technology transfer service system, such as... Figure 1 As shown, it includes: a multi-source data processing module, an analysis and evaluation module, a resource matching module, a visualization module, and a full-process management and control module, which supports the entire process from data collection, analysis, and evaluation of scientific and technological achievements to final commercial application; like Figure 2 As shown, in this embodiment, the multi-source data processing module uses a data acquisition component to collect multi-source data using acquisition tools, and a data standardization component to standardize and clean the collected multi-source data. The data acquisition component connects to the internal database of the innovation entity and the enterprise demand platform via API, and uses a distributed crawler framework to crawl public databases. The data standardization component uses a natural language processing (NLP) component to process the collected unstructured data. Specifically, a distributed crawler framework (Scrapy) is used to crawl public databases (WIPO, PubMed), and an API is used to connect to the internal database of the innovation entity (such as the laboratory management system ELN). Natural language processing components and their technologies are used to clean the unstructured data, which is then used to integrate scientific research data, market data, and collaborative resource data of scientific and technological achievements. The analysis and evaluation module constructs an evaluation model with input parameters including patent citations, experiment success rate, and market demand growth rate. Based on knowledge graph association technology and industry demand nodes, it generates a visualized technology-market mapping relationship to quantify the technological maturity, market fit, and risk assessment of scientific and technological achievements. The resource matching module uses a collaborative filtering algorithm (Surprise library) to calculate enterprise matching degree and defines matching logic through a rule engine to intelligently recommend cooperative enterprises, investors, and transformation paths for scientific and technological achievements. The visualization module includes a front-end and a back-end. The front-end uses React+D3.js to generate interactive charts (such as technology maturity radar charts and market growth curves), while the back-end integrates Monte Carlo simulation to predict the economic benefits of different paths, dynamically displaying the evaluation results, market trends, and simulated transformation path revenue and risk of scientific and technological achievements. The full-process management module manages multi-stage task dependencies through a workflow engine (Apache Airflow) and ensures data immutability by storing key nodes through blockchain (Hyperledger Fabric), tracking the transformation progress of scientific and technological achievements and automatically generating compliance documents for scientific and technological achievements. This embodiment uses multimodal data fusion to map the relationship data of technology data, market demand, and resource networks into a unified semantic space; it uses an evaluation model to generate risk scores in real time by combining technology maturity levels, patent layout, and market competitors; and it uses an intelligent matching engine that is driven by both collaborative filtering and rule engine to improve the accuracy of resource matching.

[0025] Furthermore, the multi-source data processing module connects to the internal databases of innovation entities and enterprise demand platforms via API. It cleans unstructured data (such as technical documents and semantic parsing of patent abstracts) through NLP, integrates scientific research data (patents, papers, experimental data), market data (industry demand, competitor analysis, industry reports, enterprise demand), intellectual property data (patent status, legal risks), and cooperative resources (enterprises, investors), and performs data standardization processing (unified format, deduplication, and completion). The analysis and evaluation module constructs evaluation models (based on random forest / deep learning) for technology maturity assessment, market fit analysis, and risk assessment. Technology maturity assessment, based on the TRL model, quantifies the feasibility of scientific and technological achievements. Market fit analysis uses machine learning to predict the commercialization potential of technologies (such as market size and competitive barriers). Risk assessment is used to identify intellectual property risks (infringement probability) and technological bottlenecks (such as mass production difficulty). The resource matching module recommends matching resources through collaborative filtering algorithms and defines matching logic through a rule engine (such as matching enterprise technology needs keywords with patent abstracts and keyword weight allocation) to intelligently match scientific and technological achievements with potential cooperative enterprises, investors or incubators, and generate customized transformation path suggestions (such as licensing, joint development, and startup incubation). The visualization module's backend provides a RESTful API via Python Flask, which is used to provide interactive dashboards to display technology evaluation results, market trends, resource matching degrees, and support dynamic simulation of the benefits and risks of different conversion paths (such as Monte Carlo simulation). The full-process management module is used to track the progress of technology transfer (such as cooperation negotiation, contract signing, pilot testing, and mass production), and automatically generate technology transfer reports and compliance documents (such as technology licensing contract templates).

[0026] Specifically, such as Figure 3As shown, in order to integrate multi-source heterogeneous data, unify the format, and clean the data to provide high-quality input for subsequent analysis, the multi-source data acquisition component and data standardization component of the multi-source data processing module are described in detail: The sources of multi-source data include: scientific research data, market data, and resource data; scientific research data includes: patent databases (WIPO (World Intellectual Property Organization), Derwent (Derwent Innovation Patent Analysis Database)), paper databases (PubMed (a free biomedical literature retrieval platform under the National Library of Medicine (NLM), IEEE Xplore (academic literature database of the Institute of Electrical and Electronics Engineers (IEEE) and the Institute of Engineering and Technology (IET))), and laboratory management systems (ELN (Electronic Lab Notebook, a digital tool for recording, storing, and managing laboratory experimental data)); market data includes: industry reports (such as Statista (a comprehensive global database that provides data on major countries and economies worldwide), Mordor...). Intelligence (a market research and industry analysis firm that primarily provides data-driven decision support services to enterprises), InnoCentive (a research crowdsourcing platform founded by Eli Lilly and Company, dedicated to solving complex technical problems by connecting enterprise R&D needs with global professionals)); resource data includes: business directories (such as Crunchbase (a US business information platform that includes information on company background, funding, executive teams, key personnel, competitors, etc.)), investor directories (such as PitchBook (a provider of global financial data, research, and insights)); collection tools include: web crawlers and API integration; the web crawler uses the Scrapy framework to scrape public data, and sample code is as follows: Python import scrapy class PatentSpider(scrapy.Spider): name = 'wipo_patent' start_urls = ['https: / / patentscope.wipo.int'] def parse(self, response): for patent in response.css('.patent-result'): yield { 'title': patent.css('.title::text').get(), 'abstract': patent.css('.abstract::text').get(), } API integration uses the Requests library to call third-party APIs (example code below): Python import requests def get_market_data(api_key, keyword): response = requests.get(f"https: / / api.statista.com / search?q={keyword}&apikey={api_key}") return response.json(); Data standardization and cleaning include: data format unification and data cleaning; data format unification includes: converting patent data to JSON format: {"patent_id": "US202100000A1", "claims": ["..."]}; converting experimental data to CSV format: material, efficiency, cost; PLA, 85%, 0.5$ / kg; data cleaning includes: deduplication and filling in missing values; deduplication is based on unique identifiers (such as patent numbers) using Pandas, with example code as follows: Python import pandas as pd df = pd.read_csv('raw_data.csv').drop_duplicates (subset='patent_id') The following is an example code for using KNN interpolation (Scikit-learn) to complete missing values: Python from sklearn.impute import KNNImputer imputer = KNNImputer(n_neighbors=3) df_filled = imputer.fit_transform(df); It should be noted that KNN (K-nearest neighbor) is a supervised machine learning algorithm (K-nearest neighbor algorithm). In the feature space, if the majority of the k nearest (i.e., the closest) samples in the feature space belong to a certain class, then the sample also belongs to that class. Given a training dataset, for a new input instance, find the k nearest instances (i.e., the K neighbors mentioned above) in the training dataset. If the majority of these K instances belong to a certain class, then classify the input instance into that class. The classification of a new input instance is determined based on the class of its K nearest instances. KNN interpolation, as a method for filling in missing values, uses a distance parameter, also known as the k parameter. Missing values ​​are imputed by referring to the neighborhood points of the missing value.

[0027] Further, see also Figure 3 To analyze, evaluate, and assess the technological maturity, market potential, and risks of quantitative scientific and technological achievements, and to provide a basis for decision-making, this paper provides a detailed explanation of the technology maturity assessment, market matching analysis, and knowledge graph construction modules in the analysis and evaluation module: The technology maturity assessment adopts the TRL model, and the TRL levels are defined as follows: TRL implementation uses a rule engine, which automatically makes decisions based on experimental data. Example code is as follows: Python def calculate_trl(prototype_status, certification): if certification == 'ISO-9001': return 9 elif prototype_status == 'lab_scale': return 6 else: return 3; Market fit analysis employs a machine learning model. The application of this model includes input features and model training. Input features include: patent citation count, TF-IDF values ​​of technical keywords, and market demand growth rate. Model training uses a random forest to predict commercial potential (Scikit-learn). Example code is shown below: Python from sklearn.ensemble import RandomForestRegressor X_train = df[['citation_count', 'tfidf_score', 'market_growth']] y_train = df['commercial_score'] # Manually labeled commercial score (0-100) model = RandomForestRegressor().fit(X_train, y_train); The knowledge graph was constructed using the Neo4j graph database, with nodes and relationships defined. Nodes include: Technology, Company, Market; relationships include: APPLIES_TO, COMPETES_WITH. Example code is as follows: cypher MATCH (t:Technology {name: "Degradable Nanomaterials"})-[r:APPLIES_TO]->(m:Market) RETURN m.name, r.strength; Further, see also Figure 3 To intelligently match technology with businesses / investors and generate conversion paths, this document provides a detailed explanation of the collaborative filtering algorithm and rule engine (Drools) within the resource matching module. The implementation steps of the collaborative filtering algorithm include: constructing a "business-technology" matrix, where the matrix values ​​are matching scores; and implementing collaborative filtering using the Surprise library, with example code as follows: Python from surprise import SVD, Dataset, Reader reader = Reader(rating_scale=(0, 100)) data = Dataset.load_from_df(df[['company_id', 'tech_id', 'score']],reader) algo = SVD().fit(data.build_full_trainset()) The following is an example of rule code from the rule engine (Drools): drools Rule "Matching Environmentally Friendly Enterprises" when $t: Technology (keywords include "biodegradable") $c: Company(industry == "Environmental Protection") then insert(new Match($t, $c, 90)); end Further, see also Figure 3 To provide an interactive Kanban board and support dynamic simulation, a detailed explanation of the front-end implementation of the visualization module and the Monte Carlo simulation is provided: The example code for the front-end implementation (React + D3.js) is as follows: javascript import { BarChart} from 'd3-components'; function MarketTrendChart({ data}) { return <barchart data="{data}" x="year" y="growth_rate" / > ; }; The following is an example code for a risk-reward simulation using Monte Carlo simulation: Python import numpy as np def monte_carlo_simulation(expected_revenue, risk_score, iterations=1000): revenues = [] for _ in range(iterations): noise = np.random.normal(0, risk_score * 0.1) revenues.append(expected_revenue * (1 + noise)) return np.percentile(revenues, [5, 50, 95]); Further, see also Figure 3 To manage conversion progress and ensure compliance and traceability, a detailed explanation of the workflow engine (Apache Airflow) and blockchain notarization (Hyperledger Fabric) of the end-to-end control module is provided: Example code for the DAG definition of the workflow engine (Apache Airflow) is as follows: Python import DAG from airflow from airflow.operators.python import PythonOperator dag = DAG('tech_transfer', schedule_interval='@weekly') task1 = PythonOperator( task_id='data_collection', python_callable=run_spider, dag=dag ); The following is a sample code for a smart contract (Chaincode) in Hyperledger Fabric: go func (s *SmartContract) RecordProgress(ctxcontractapi.TransactionContextInterface, techIDstring, stage string) error {data := map[string] string{"stage": stage, "timestamp": time.Now().String()} return ctx.GetStub().PutState(techID, []byte(data)) }; In practical applications of this embodiment, issues such as data silos, subjective evaluation, and compliance risks may arise. For data silos, federated learning can be used for joint modeling without sharing the original data. For subjective evaluation, a multi-expert scoring mechanism can be introduced to calibrate the model output. For compliance risks, a legal knowledge base (such as GDPR and patent law) can be integrated to automatically generate compliance recommendations. Federated learning uses the PySyft library, with each institution training the model locally and sharing only the model parameters. The multi-expert voting mechanism in the expert scoring system integrates a weighted voting mechanism to calibrate the model output. The legal knowledge base uses a rule base to automatically check contract terms. Example code is as follows: Python def check_contract(clause): if "exclusive_license" in clause and "territory:Asia" not inclause: The risk is not specified in terms of geographical scope. Based on the above, the technology transfer service system of this embodiment can achieve full-chain coverage, dynamic optimization, and transparency and trustworthiness, realizing seamless connection of each link from data collection to commercialization. It continuously improves the accuracy of evaluation and matching through machine learning, and blockchain storage ensures that the transfer process is traceable and tamper-proof. The system significantly improves the efficiency and success rate of technology transfer through intelligent toolchains, and is suitable for universities, research institutions and enterprise technology transfer centers.

[0028] For example, based on this embodiment, the entire process of patent commercialization in universities can be as follows: data collection, such as crawling patent USxxxxxxxxxxx, extracting claims and experimental data; technology evaluation, TRL score of 4 (laboratory verification), market matching score of 85; resource matching, recommending company A (matching score of 92%), generating a joint development proposal; commercialization execution, Airflow triggers contract generation task, blockchain records signing nodes; progress tracking, uploading pilot-scale data, system warning of cost overruns, and automatic push of process optimization solutions; specifically, in the patent commercialization scenario of universities, for example, during data collection, crawling a university's "degradable nanomaterials" patent (US1), extracting claims and experimental data (degradation efficiency 85%, cost $0.5 / kg); during intelligent evaluation, TRL... The assessment level is 4 (laboratory verification stage), with a market matching score of 85 (environmental protection industry demand is increasing by 15% annually). During resource matching, Company A (92% matching) is recommended, and a joint development proposal is generated. During implementation, Airflow triggers a contract generation task, the blockchain records the signing node, data is uploaded in real-time during the pilot-scale stage, the system warns of a 20% cost overrun, and pushes process optimization solutions. In the technology transfer scenario of medical institutions, such as the implementation process of an "AI-assisted diagnostic algorithm" in a hospital: during data collection, algorithm performance data (accuracy 95%) and medical industry regulations (FDA certification requirements) are integrated; during risk assessment, data privacy risks (HIPAA compliance) are identified, with a score of 65; during path matching, cooperation with medical AI companies is recommended, generating a phased commercialization plan.

[0029] The intelligent technology transfer service system in this embodiment achieves full-process management from technology evaluation to commercialization through multi-source data fusion, dynamic risk assessment and intelligent matching engine. The system innovatively combines TRL model, knowledge graph and blockchain technology to significantly improve the efficiency and success rate of transformation. It is suitable for technology transfer scenarios in universities, medical institutions, research institutions and enterprises.

[0030] Example 2: This example provides a data processing method for technology transfer, applied to the technology transfer service system of Example 1, such as... Figure 4 As shown, it includes: S10: Perform data fusion and standardization processing on scientific research data, market data, intellectual property data, and cooperative resource data; S20: By constructing an evaluation model, the technology maturity assessment, market fit analysis, and risk assessment are conducted on the integrated and standardized data. S30: Based on the evaluation and analysis results, the system intelligently matches scientific and technological achievements with potential partner companies, investors or incubators through collaborative filtering algorithms and rule engines, and generates customized transformation path suggestions. In step S10, scientific research data, market data, intellectual property data, and cooperative resource data are fused and standardized to map data from different sources (such as technical parameters and market demand) to the same semantic space. Specifically, this involves NLP technology and feature engineering to vectorize and align keywords in patent texts and industry reports. In step S20, the inputs to the evaluation model include TRL, patent coverage and competitor technology layout, and the output includes risk score, which mainly involves the impact of different weights and the training process of the model, such as the algorithm used, data sources and verification methods. In step S30, intelligent matching includes collaborative filtering and a rule engine; collaborative filtering includes building a user-technology matrix to handle cold starts; the rule engine includes keyword matching logic to improve matching accuracy.

[0031] Furthermore, in step S10, data fusion and standardization processing are performed on research data, market data, intellectual property data, and cooperative resource data, such as... Figure 5 As shown, it mainly includes: S101: Cross-modal data alignment; S102: Active learning annotation; Cross-modal data alignment is used to map technical parameters (such as material properties) to market demands (such as industry pain points) into a unified semantic space; Actively learn annotation and optimize the evaluation model through expert feedback (such as annotating high-potential technology cases). The assessment model takes Technology Readiness Level (TRL), patent family coverage, and competitor technology layout as inputs, and outputs a risk score (0-100 points), for example: Risk Score=0.4×Legal Risk+0.3×Technical Risk+0.3×Market Risk; Intelligent matching includes: calculating the enterprise-technology matching degree, and the calculation formula is as follows: ; in, Technical keywords (such as "nanomaterials" and "biodegradation") Industry weighting (e.g., the environmental protection industry has a higher weighting).

[0032] Specifically, in step S10, data fusion and standardization are performed on scientific research data, market data, intellectual property data and cooperative resource data. This is mainly achieved through cross-modal data alignment in step S101 and active learning annotation in step S102, which integrates multi-source heterogeneous data (technical parameters, market demand, resource network) to improve data consistency and information density. Step S101 Cross-modal data alignment is achieved through techniques such as semantic embedding (NLP), knowledge graph alignment, and loss function design; Semantic embedding (NLP) uses pre-trained models (such as BERT) to convert textual data (patent abstracts, industry reports) into vectors. Example code is shown below: Python from transformers import BertTokenizer, BertModel tokenizer = BertTokenizer.from_pretrained('bert-base-uncased') model = BertModel.from_pretrained('bert-base-uncased') text = "Degradable nanomaterials for eco-friendly packaging" inputs = tokenizer(text, return_tensors='pt') outputs = model(**inputs) embedding = outputs.last_hidden_state.mean(dim=1) # 768-dimensional vector; The structured data (experimental parameters, market size values) are standardized (Z-score) and then concatenated into the same vector space; Knowledge graph alignment is achieved by defining cross-modal node relationships in Neo4j. Example code is shown below: cypher / / Linking technology nodes with market nodes MATCH (t:Technology {name: "Degradable Nanomaterials"}), (m:Market {name: "Eco-friendly Packaging"}) MERGE (t)-[r:APPLIES_TO {score: 0.85}]->(m); The loss function is designed using contrastive learning to minimize the distance between similar data and maximize the distance between dissimilar data, as shown in the following formula: ; Where α is the interval hyperparameter (e.g., 1.0); Step S102, the process of active learning and labeling, includes: the initial model is trained using a small amount of labeled data (such as 100 technical-market matching samples); the model predicts on unlabeled data and selects the sample with the highest uncertainty (such as a prediction probability close to 0.5); after expert labeling, the sample is added to the training set, and the model is iteratively optimized. The implementation of active learning annotation (active learning loop), with example code as follows: Python from modAL.uncertainty import entropy_sampling from sklearn.ensemble import RandomForestClassifier # Initial training set X_labeled, y_labeled = load_initial_data() model = RandomForestClassifier().fit(X_labeled, y_labeled) # Active Learning Cycle for _ in range(10): # Iterate 10 times X_pool = load_unlabeled_data() query_idx = entropy_sampling(model, X_pool)X_query, y_query = expert_label(X_pool[query_idx]) model.fit(np.vstack([X_labeled, X_query]), np.concatenate([y_labeled, y_query])).

[0033] In step S20, an evaluation model is constructed to conduct technology maturity assessment, market fit analysis, and risk assessment on the fused and standardized data. The application of the evaluation model is as follows: Figure 6As shown, the main steps are: input parameters and their calculation in step S201, risk scoring in step S202, and model training in step S203. The risk of technology transformation is quantified in real time, and the weights and thresholds are dynamically adjusted. The input parameters and calculations for step S201 include: Technology Readiness Level (TRL), patent coverage, and competitor technology layout; Technology maturity is determined based on the experimental stage (levels 1-9), with example code as follows: Python def get_trl(prototype_status, certification): if certification in ['ISO-9001', 'FDA']: return 9 elif prototype_status == 'pilot_scale': return 7 else: return max(1, int(experiment_success_rate * 9)) # Map the experiment success rate to TRL; Patent coverage is used to calculate the number of countries covered by a patent family (e.g., PCT applications covering 50 countries → coverage = 50 / 195 ≈ 25.6%). Competitor technology layouts are analyzed using patent citation networks to calculate technology similarity (Cosine similarity): ; Among them, Similar Patents represents the number of similar patents, and Total Patents in Domain represents the total number of patents in the domain; The formula for calculating the risk score in step S202 is as follows: ;

[0034] Weight allocation, for example: w1=0.4, w2=0.3, w3=0.3; Sub-item calculation: TRL Risk: (The lower the TRL, the higher the risk); Legal Risk: ;

[0035] Market Risk: ;

[0036] Step S203, model training, includes: training data and model selection; the training data source is historical conversion project data (features are TRL, patent coverage, number of competitors; labels are success / failure); the model selection is Gradient Boosting Tree (XGBoost), which supports feature importance analysis; the code example is as follows: Python import xgboost as xgb from sklearn.model_selection import train_test_split X, y = load_risk_data() X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2) model = xgb.XGBClassifier().fit(X_train, y_train) # Feature Importance Visualization xgb.plot_importance(model); In step S30, based on the evaluation and analysis results, the collaborative filtering algorithm and rule engine intelligently match technological achievements with potential partner companies, investors, or incubators, generating customized transformation path suggestions. The intelligent matching process is as follows: Figure 7 As shown, it mainly uses the application of S301 collaborative filtering (user-technology matrix), S302 rule engine (Drools) and S303 hybrid matching to accurately match scientific and technological achievements with enterprise needs and support the integration of multiple strategies. The collaborative filtering in step S301 includes matrix construction and matrix completion; the matrix construction table is as follows: Technical scores are derived from historical collaboration data, manual annotation, and automatic calculations (such as keyword matching degree). Matrix completion uses SVD decomposition to predict missing values: ; Where, q i p is the technology latent vector. u For enterprise latent vectors; code implementation (Surprise library) example is as follows: Python from surprise import SVD, Dataset, accuracy data = Dataset.load_from_df(df[['company_id', 'tech_id', 'score']],reader) trainset = data.build_full_trainset() algo = SVD(n_factors=50).fit(trainset) # Predict Company A's rating of Technology X prediction = algo.predict('CompanyA', 'TechX') print(prediction.est) # Output the predicted rating; The application of the rule engine in step S302 mainly includes: rule definition of the rule engine (Drools) and execution of the rule engine; The rules of the rule engine (Drools) are defined using the following code: drools The rule "matches high-growth companies" when $t: Technology(keywords contains "AI Medical", trl>= 7) $c: Company(industry == "Healthcare", growth_rate>20%) then insert(new Match($t, $c, 95)); End; The execution flow of the rule engine in step S302 is as follows: Figure 8 As shown, it includes: S3021: Input technical and enterprise characteristics into the rule engine; S3022: Trigger all matching rules and generate a candidate matching list; S3023: Sort by rating and output Top-N recommendation results; The hybrid matching strategy in step S303 includes: weighted fusion and cold start processing; The weighted fusion formula is as follows: ;

[0037] Cold start processing includes: when a new enterprise / technology has no historical data, relying solely on the rule engine and keyword matching; In the actual implementation of this embodiment, verification and optimization are still required, including: data fusion verification, evaluation model verification, and matching algorithm optimization. The data fusion verification metric is the cross-modal alignment accuracy (e.g., whether the technology-market association is correct), and the method is to manually sample 100 alignment results and calculate an accuracy ≥ 90%. The evaluation model verification uses ROC curves and AUC values, and the implementation code example is as follows: Python from sklearn.metrics import roc_curve, auc y_pred = model.predict_proba(X_test)[:, 1] fpr, tpr, _ = roc_curve(y_test, y_pred) print("AUC:", auc(fpr, tpr)) # Target AUC > 0.85; The matching algorithm optimization adopted A / B testing, including a control group and an experimental group. The control group used traditional keyword retrieval, while the experimental group used a hybrid matching strategy. The indicator was an improvement of matching success rate (actual cooperation rate) of ≥30%.

[0038] The technology transfer data processing method in this embodiment achieves a closed loop from data to decision-making and full-process management from technology evaluation to commercialization. It innovatively combines TRL models, knowledge graphs, and blockchain technology to significantly improve the efficiency and success rate of technology transfer. It is applicable to technology transfer scenarios in universities, research institutions, and enterprises.

[0039] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the prior art, can be embodied in the form of software products. These computer software products can be stored in computer-readable storage media, such as ROM / RAM, magnetic disks, optical disks, etc., and include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute certain parts of the methods described in the embodiments. It should be noted that if this solution uses terms such as module, component, part, or component, it should be interpreted broadly in context; it can refer to hardware, software, or a functional part of hardware and software combined or associated. The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them.

Claims

1. A technology transfer service system, characterized in that, include: The multi-source data processing module collects multi-source data through a data acquisition component and standardizes and cleans the multi-source data through a data standardization component. The analysis and evaluation module generates a visualized technology-market mapping relationship by constructing an evaluation model based on a knowledge graph and industry demand nodes. The resource matching module uses collaborative filtering algorithm technology to calculate enterprise matching degree, and defines matching logic through a rule engine. The visualization module includes a front-end and a back-end. The front-end generates interactive charts, and the back-end integrates Monte Carlo simulation to predict economic benefits. The end-to-end management module manages multi-stage task dependencies through a workflow engine and stores data using blockchain. The input parameters of the evaluation model include the number of patent citations, the success rate of experiments, and the market demand growth rate.

2. The technology transfer service system according to claim 1, characterized in that: The data acquisition component connects to the internal database of the innovation entity and the enterprise demand platform via API; The analysis and evaluation module uses an evaluation model to assess technology maturity, analyze market fit, and assess risks. The technology maturity assessment is based on the TRL model to quantify the feasibility of scientific and technological achievements; The market fit analysis uses machine learning to predict the commercial potential of the technology. The risk assessment is used to identify intellectual property risks and technological bottlenecks; The resource matching module uses a collaborative filtering algorithm to recommend and match scientific and technological achievements with potential partners, investors, or incubators, and generates transformation path suggestions. The backend provides an interactive dashboard that displays technology evaluation results, market trends, and resource matching, and supports dynamic simulation of the benefits and risks of different conversion paths.

3. The technology transfer service system according to claim 2, characterized in that: The data acquisition component uses a distributed crawler framework to crawl public databases and obtain unstructured data; The data standardization component processes the collected unstructured data using a natural language processing component; The multi-source data includes: scientific research data, market data, and resource data; The research data includes: patent database, paper database, and laboratory management system; The market data includes: industry reports and business demand; The resource data includes: business directories and investor lists; The cleaning process includes: deduplication and filling in missing values; The deduplication is performed using Pandas based on a unique identifier; The missing values ​​are filled using KNN interpolation.

4. The technology transfer service system according to claim 3, characterized in that: The analysis and evaluation module is used for technology maturity assessment, market fit analysis, and knowledge graph construction. The technology maturity assessment uses the TRL model to determine the technology maturity level based on the experimental stage. The TRL model includes a rule engine, which automatically makes decisions based on experimental data. The market matching analysis employs a machine learning model, including: input features and model training; The input features include: patent citation count, TF-IDF value, and market demand growth rate; The model was trained using a random forest to predict commercial potential. The knowledge graph is constructed using a knowledge graph database tool to define nodes and relationships.

5. The technology transfer service system according to claim 2, characterized in that: The implementation steps of the collaborative filtering algorithm include: Construct an enterprise-technology matrix, with matrix values ​​representing a matching score; Implement collaborative filtering using the Surprise library.

6. A method for processing data related to the transformation of scientific and technological achievements, applied to the scientific and technological achievement transformation service system described in any one of claims 1-5, characterized in that, include: Acquire scientific research data, market data, intellectual property data, and resource data, and perform data fusion and standardization processing; By constructing an evaluation model, we can conduct technology maturity assessment, market fit analysis, and risk assessment on the data after fusion and standardization. Based on the evaluation and analysis results, the system intelligently matches scientific and technological achievements with potential partner companies, investors, or incubators through collaborative filtering algorithms and rule engines, generating transformation path suggestions.

7. The data processing method for the transformation of scientific and technological achievements according to claim 6, characterized in that: The acquisition of scientific research data, market data, intellectual property data, and resource data, and the subsequent data fusion and standardization processing, include: Connect to the internal databases of innovation entities and enterprise demand platforms via API; Clean unstructured data using natural language processing components; Integrate scientific research data, market data, intellectual property data, and resource data; Perform data standardization processing; The process of constructing an evaluation model to assess the technology maturity, market fit, and risk of the fused and standardized data includes: Technology readiness assessment is based on the TRL model to quantify the feasibility of scientific and technological achievements; Market fit analysis uses machine learning to predict the commercial potential of technologies; Risk assessment identifies intellectual property risks and technological bottlenecks.

8. The data processing method for the transformation of scientific and technological achievements according to claim 7, characterized in that: The data fusion and standardization processing includes: Mapping data from different sources to the same semantic space; Vectorize and align keywords in patent texts and industry reports; The assessment model takes into account technology maturity, patent coverage, and competitor technology layout as inputs, and outputs a risk score.

9. The data processing method for the transformation of scientific and technological achievements according to claim 8, characterized in that: The data fusion and standardization processing includes: cross-modal data alignment and active learning annotation; The cross-modal data alignment includes: semantic embedding and knowledge graph alignment; the semantic embedding includes: converting text data into vectors using a pre-trained model; and standardizing structured data and concatenating them into the same vector; The knowledge graph alignment includes: defining cross-modal node relationships in the knowledge graph; The active learning annotations optimize the evaluation model through expert feedback. The technology maturity level is determined based on the experimental stage. The patent coverage calculation measures the number of countries covered by the patent family. The competitor's technology layout includes: calculating technology similarity through a patent citation network.

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, it implements the technology transfer service system according to any one of claims 1 to 5.

Images