Intelligent construction system for enterprise patent database

Abstract

The invention provides an intelligent construction system for an enterprise patent database. The intelligent construction system comprises a data acquisition module, a cross-industry technology identification module, a technology cross optimization module and a result presentation and recommendation module, the data acquisition module is used for acquiring patent data from patent databases of different industries and ensuring the comprehensiveness of cross-industry patent data in combination with internal research and development data of enterprises; the cross-industry technology identification module performs semantic analysis on patent data in different fields based on natural language processing and machine learning technologies, identifies potential technology intersections and constructs a cross-industry technology similarity matrix; the technology cross optimization module optimizes technology cross points, combines innovativeness scores, application breadth and market potential dimensions of the technology, evaluates feasibility of technology fusion and recommends an optimal technology combination; and the result presentation and recommendation module adopts a visual tool to display technical intersections and potential application scenes, and recommends related patent combinations and technical innovation schemes to the user.

Description

Technical Field

[0001] The present invention relates to the field of patent data fusion technology, and in particular to an enterprise patent database intelligent construction system. Background Art

[0002] Patent cross-industry integration technology is an important means to promote industrial upgrading and the formation of emerging industries. Many companies have begun to build patent databases for their own industries, which are often limited to the data and technology of a single industry. It is difficult to identify the potential for cross-industry technology application, causing companies to miss many opportunities for technology integration when innovating.

[0003] In addition, despite the high output of patents, there is a gap in the proportion of high-quality patents, and the transformation implementation is not ideal, resulting in the failure to form a cross-industry patent integration technology service system and an inefficient operating mechanism. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides an intelligent construction system for an enterprise patent database. By building a tool that can identify technical intersections between different fields, the system can automatically identify potential cross-industry technical intersections through machine learning and patent text analysis technology, and prompt enterprises to bring about technical integration opportunities that patents in different industries may bring. This cross-identification can not only help enterprises to carry out cross-border innovation, but also bring new opportunities for technology transfer and patent authorization.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions: An intelligent construction system for enterprise patent database, the system includes the following modules: Data collection module: used to collect patent data from patent databases of different industries, including patent text, images, citation data, and combine with internal R&D data of enterprises to ensure the comprehensiveness of cross-industry patent data; Cross-industry technology identification module: Based on natural language processing and machine learning technology, it conducts semantic analysis on patent data in different fields, identifies potential technology intersections, and builds a cross-industry technology similarity matrix; Technology cross-optimization module: By optimizing technology cross-points, combined with the innovation score, application breadth, and market potential of the technology, the feasibility of technology integration is evaluated and the optimal technology combination is recommended; Result presentation and recommendation module: Use visualization tools to display technology intersections and potential application scenarios, and recommend relevant patent portfolios and technological innovation solutions to users.

[0006] Preferably, the data acquisition module can regularly obtain the latest patent data from major patent offices around the world, support real-time updates, and standardize the data to ensure that the data formats from different sources are consistent.

[0007] Preferably, the cross-industry technology identification module identifies cross-industry technology intersections in the following manner: Extract key technical features from patent texts through semantic analysis technology; Use TF-IDF algorithm and deep learning model to calculate the technical similarity of patents in different industries; Generate a cross-industry technology similarity matrix and identify technology intersections based on clustering algorithms.

[0008] Preferably, the cross-industry technology similarity matrix is ​​generated by calculating the cosine similarity between patent technology feature vectors.

[0009] Preferably, the clustering algorithm uses K-means algorithm to perform patent clustering, and the optimization goal is to minimize the differences in technical features within the cluster to identify the technology combination with the greatest integration potential.

[0010] Preferably, the technology cross optimization module performs technology combination optimization by genetic algorithm and evaluates the fusion potential of technology combination by fitness function, and the fitness function includes the following parameters: Innovation rating of the technology; The breadth of application of the technology; Market potential; Number of patent citations.

[0011] Preferably, the genetic algorithm comprises the following steps: 1) Encoding patent technical features into gene sequences; 2) Select the best technology combination through the fitness function for cross-breeding and mutation operations; 3) Iterate the optimization until the fitness function converges and output the optimal technology combination.

[0012] Preferably, the result presentation and recommendation module displays cross-industry technology intersections in the form of a graph or heat map, and intelligently recommends the optimal technology combination based on the evaluation results of technology integration potential.

[0013] Preferably, the system is able to dynamically monitor the technological development trends of different industries, regularly update the patent database, and ensure that the analysis results of technology intersections are synchronized with the latest technological developments.

[0014] Compared with the prior art, the present invention has the following beneficial effects: This invention builds a tool that can identify technology intersections between different fields. Through machine learning and patent text analysis technology, the system can automatically identify potential cross-industry technology intersections and prompt companies to identify technology integration opportunities that may be brought about by patents in different industries. This cross-identification can not only help companies conduct cross-border innovation, but also bring new opportunities for technology transfer and patent licensing.

[0015] Based on the technological development trajectories of different industries, the present invention can predict the cross-industry technological integration trends that may appear in the future, and recommend relevant patent portfolios for reference by enterprises, thereby guiding enterprises to deploy forward-looking innovative technologies. DETAILED DESCRIPTION

[0016] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art can think of other obvious variations. The basic principles of the present invention defined in the following description can be used for other embodiments, variations, improvements, equivalents, and other technical solutions that do not deviate from the spirit and scope of the present invention.

[0017] The present invention provides an enterprise patent database intelligent construction system, comprising the following modules: Data collection module: used to collect patent data from patent databases of different industries, including patent text, images, citation data, and combine with internal R&D data of enterprises to ensure the comprehensiveness of cross-industry patent data; Cross-industry technology identification module: Based on natural language processing and machine learning technology, it conducts semantic analysis on patent data in different fields, identifies potential technology intersections, and builds a cross-industry technology similarity matrix; Technology cross-optimization module: By optimizing technology cross-points, combined with the innovation score, application breadth, and market potential of the technology, the feasibility of technology integration is evaluated and the optimal technology combination is recommended; Result presentation and recommendation module: Use visualization tools to display technology intersections and potential application scenarios, and recommend relevant patent portfolios and technological innovation solutions to users.

[0018] The data acquisition module of the present application is used for data acquisition and preprocessing.

[0019] Data collection includes: obtaining patent data from different industry patent databases (such as medical, communications, materials science, etc.), covering the data resources of major patent offices around the world. Combining internal R&D data of enterprises, such as technical reports, innovation proposals, technical solutions, etc., to enrich the basic data for cross-identification.

[0020] Data collection preprocessing includes data standardization and patent classification and annotation; among them, Data standardization: Data from different sources are standardized to ensure consistent data formats for easy subsequent analysis and matching.

[0021] Patent classification and annotation: Annotate patent data according to industry fields and technical classifications to form a basic classification library of industry patents.

[0022] The cross-industry technology identification module of this application is based on natural language processing and machine learning technology to perform semantic analysis on patent data in different fields, identify potential technology intersections, and construct a cross-industry technology similarity matrix. Specifically,

[0023] Semantic analysis and technical similarity calculation: Use natural language processing technology to deeply analyze the technical descriptions and innovations in patent texts and extract key technical points; By using the method of technical feature quantification, the technical similarities between patents in different industries are calculated to identify patent technologies with potential cross-applications.

[0024] Cross-domain technical links: Combine citation data, patent holders’ backgrounds, technology application fields and other multi-dimensional information to analyze the cross-domain relevance of technology. Through deep learning model training, the system can identify similar or complementary technologies used in different industries;

[0025] Construct a cross-industry technology correlation map to display the correlation structure between patents and the cross-industry technology overlap areas.

[0026] The purpose of this cross-industry technology identification module is to discover cross-industry technology integration opportunities by extracting technology features from patent databases of multiple industries and identifying potential connections between different technologies. The key steps are as follows:

[0027] 1.1 Semantic analysis and technical feature extraction of patent text Text preprocessing: 1. Remove stop words: First, preprocess the patent text and remove unimportant stop words (such as "the", "and", etc.) to reduce the interference of irrelevant information on semantic analysis.

[0028] 2. Stemming: Stemming words and merging different forms of words, such as "develop" and "developing".

[0029] Technical feature extraction algorithm: Use the TF-IDF (term frequency-inverse document frequency) algorithm to quantify the technical features of patent texts: A word vector is generated for each patent document to indicate the importance (weight) of each word.

[0030] Calculate the TF-IDF value for each word:

[0031] in: TF(t,d) is the frequency of word t in document d, N is the total number of all patent documents, and DF(t) is the number of documents containing word t. Each patent is vectorized by TF-IDF to generate a technical feature vector of the patent.

[0032] Deep semantic understanding model: Pre-trained language models such as BERT (Bidirectional Encoder Representations from Transformers) are used to understand the deep semantic structure of patent texts. BERT can capture contextual semantic associations from patent descriptions and generate more sophisticated technical feature vectors.

[0033] For each patent, BERT is used to map the text content of the patent into a high-dimensional semantic vector: Vpatent=BERT(Tpatent), where Vpatent is the semantic vector of the patent text and Tpatent is the input of the patent text.

[0034] 1.2 Calculation of cross-industry technology similarity Cosine Similarity: The similarity between different patent technologies is determined by calculating the cosine similarity between patent technology feature vectors:

[0035] Among them, V 1 and V 2 They are the feature vectors of two patents respectively, and the dot product represents the similarity of the two vectors.

[0036] Cross-industry similarity matrix: For patents in different industries, a similarity matrix is ​​generated through cosine similarity. Each element in the matrix represents the similarity between two patent technologies: Si,j=cosine_similarity(Vpatenti,Vpatentj) Among them, Si,j is the similarity value between the i-th industry patent and the j-th industry patent.

[0037] 1.3 Identification of cross-industry technology relevance Clustering Algorithm: The K-means clustering algorithm is used to cluster patents and classify patents with high similarity into one category to identify technical connections across industries.

[0038] The key steps of the K-means algorithm are: 1. Randomly select k initial cluster centers.

[0039] 2. Calculate the Euclidean distance from each patent vector to the cluster center and assign the patent to the center with the closest distance.

[0040] 3. Calculate the mean of each cluster as the new cluster center.

[0041] 4. Repeat steps 2 and 3 until the cluster center no longer changes.

[0042] Clustering optimization goal:

[0043] Among them, Vci is the center of the i-th cluster, Vpatentj is the patent vector belonging to the cluster, and the goal is to minimize the intra-cluster distance.

[0044] The similarity matrix reduces the computational burden of the optimization phase by pre-screening high-similarity technologies, and provides similarity weights during the optimization process, so that the optimization algorithm tends to favor technology combinations with greater fusion potential.

[0045] The clustering optimization goal is to cluster similar technologies to provide a modular optimization structure, improve optimization efficiency and combination success rate. Clustering also provides a hierarchical processing method for technology cross-optimization, helping enterprises better manage the technology combination process.

[0046] The main goal of the technology cross-optimization module of this application is to evaluate the integration potential of cross-industry technologies and recommend innovations with the greatest market and technical value. Including:

[0047] Cross-application scenario mining: Combining the technical characteristics of patents with market demand, we can identify possible scenarios for cross-industry applications of technologies. For example, if a certain material technology has potential for application in both the medical and automotive fields, the system will automatically explore these potential market applications.

[0048] Technology Convergence Potential Assessment: By analyzing the innovation of technology, the breadth of application scenarios and market demand, we evaluate the feasibility and market potential of cross-industry technology integration. We also establish an optimization model to help companies evaluate which technology intersections are most valuable.

[0049] Technology portfolio optimization: Combined with the technologies in the patent pool, the system can recommend which existing technology combinations have advantages in the integration of different industries and optimize the company's R&D layout. For example, the system can automatically generate solutions for technology combinations in different fields to help companies simultaneously lay out innovations in the medical and electronics industries.

[0050] The specific steps and algorithms of the cross-optimization module of this technology are as follows: 2.1 Assessment of Technology Integration Potential Technology Convergence Potential Model: For cross-industry technology combinations, a technology integration potential assessment model is designed. This model evaluates the potential of technology crossover based on the innovation of the technology, the breadth of application scenarios, and market prospects.

[0051] Fusion Potential Formula: <![CDATA[F fusion =a 1 ·I novelty +α2·B breadth +α3·M market ]]> in: Fusion Potential Formula: Ffusion=α 1 ·Inovelty+α2·Bbreadth+α3·Mmarket in: Inovelty is the innovativeness score of a technology, based on factors such as the number of patent citations and patent originality.

[0052] Bbreadth is the breadth of application of a technology, measuring its applicability across multiple industries.

[0053] Mmarket is the market potential, reflecting the demand for the technology in the target market.

[0054] α1, α2, and α3 are the weights of each indicator, which are adjusted dynamically based on enterprise needs.

[0055] 2.2 Technology combination optimization Technology Combination Algorithm: In order to optimize the technology combination, a genetic algorithm (GA) is used for global search. The genetic algorithm can find the optimal technology combination in a complex search space.

[0056] Key steps: 1. Encoding: Represent the patent technology as a gene sequence in the genetic algorithm, where each gene represents the characteristics of a patent technology.

[0057] 2. Fitness function: The fitness value of each technology combination is calculated according to the technology fusion potential formula Ffusion.

[0058] 3. Selection: Select the parent of the technology combination for cross breeding based on the fitness value.

[0059] 4. Crossover and mutation: Randomly exchange some gene loci to generate new technology combinations and perform random mutation operations.

[0060] 5. Iteration: Repeat crossover, mutation, and selection operations until the fitness function converges.

[0061] Ultimately, the genetic algorithm outputs the technology combination with the highest fitness value and recommends it to the enterprise for cross-industry innovation.

[0062] The result presentation and recommendation module of this application uses visualization tools to display technology intersections and potential application scenarios, and recommends relevant patent portfolios and technological innovation solutions to users. Specifically, it includes:

[0063] Visualization: Using knowledge graphs and interactive visualization tools, we can show the technical intersections and potential application scenarios of different patents. Through the graphs, corporate R&D personnel can intuitively see which technologies overlap and have integration opportunities in different industries.

[0064] Industry technology intersection heat map: The heat map shows the dense areas of technology intersection between industries, helping companies to intuitively identify which industries have more technology integration points.

[0065] Smart Recommendation: Based on the cross-identification and optimization results, the system automatically recommends innovative directions for technological integration and potential patent combinations to enterprises and provides specific implementation plans.

[0066] The system can dynamically adjust recommended content based on the company's industry needs to ensure that technology intersections are in line with the company's business development strategy.

[0067] In addition, this application also includes a system automation and dynamic update module, which can dynamically monitor the technological development trends of different industries, regularly update the patent database, and ensure that the analysis results of technology intersections are synchronized with the latest technological trends. Specifically, it includes:

[0068] Automated technology identification: The system automatically updates patent data and internal technical documents of the enterprise on a regular basis to keep the information in the database real-time and ensure that the analysis results of technology intersections are updated at any time.

[0069] Dynamic intersection monitoring: The system automatically monitors the technological development trends in different industries, continuously tracks which fields have increasing technological crossover or emerging technologies are on the rise, and provides timely warnings and suggestions to companies.

[0070] Those skilled in the art should understand that the embodiments of the present invention shown in the above description are only examples and do not limit the present invention. The purpose of the present invention has been fully and effectively achieved. The functions and structural principles of the present invention have been demonstrated and explained in the embodiments, and the embodiments of the present invention may be deformed or modified in any way without departing from the principles.

Claims

1. An intelligent construction system for enterprise patent database, characterized by: The system includes the following modules: Data collection module: used to collect patent data from patent databases of different industries, including patent text, images, citation data, and combine with internal R&D data of enterprises to ensure the comprehensiveness of cross-industry patent data; Cross-industry technology identification module: Based on natural language processing and machine learning technology, it conducts semantic analysis on patent data in different fields, identifies potential technology intersections, and builds a cross-industry technology similarity matrix; Technology cross-optimization module: By optimizing technology cross-points, combined with the innovation score, application breadth, and market potential of the technology, the feasibility of technology integration is evaluated and the optimal technology combination is recommended; Result presentation and recommendation module: Use visualization tools to display technology intersections and potential application scenarios, and recommend relevant patent portfolios and technological innovation solutions to users.

2. According to claim 1, the enterprise patent database intelligent construction system is characterized by: The data acquisition module can regularly obtain the latest patent data from major patent offices around the world, support real-time updates, and standardize the data to ensure that the data formats from different sources are consistent.

3. According to claim 2, the enterprise patent database intelligent construction system is characterized by: The cross-industry technology identification module identifies cross-industry technology intersections in the following ways: Extract key technical features from patent texts through semantic analysis technology; Use TF-IDF algorithm and deep learning model to calculate the technical similarity of patents in different industries; Generate a cross-industry technology similarity matrix and identify technology intersections based on clustering algorithms.

4. The enterprise patent database intelligent construction system according to claim 3 is characterized by: The cross-industry technology similarity matrix is ​​generated by calculating the cosine similarity between patent technology feature vectors.

5. The enterprise patent database intelligent construction system according to claim 3 is characterized by: The clustering algorithm uses the K-means algorithm to perform patent clustering, and the optimization goal is to minimize the differences in technical features within the cluster to identify the technology combination with the greatest integration potential.

6. An enterprise patent database intelligent construction system according to any one of claims 1 to 5, characterized in that: The technology cross optimization module optimizes the technology combination through a genetic algorithm and uses a fitness function to evaluate the fusion potential of the technology combination. The fitness function includes the following parameters: Innovation rating of the technology; The breadth of application of the technology; Market potential; Number of patent citations.

7. The enterprise patent database intelligent construction system according to claim 6, characterized in that: The genetic algorithm comprises the following steps: Encoding patented technical features into gene sequences; The best technology combination is selected through the fitness function for cross-breeding and mutation operations; The optimization is performed iteratively until the fitness function converges and the optimal technology combination is output.

8. The enterprise patent database intelligent construction system according to claim 7 is characterized by: The result presentation and recommendation module displays cross-industry technology intersections in the form of graphs or heat maps, and intelligently recommends the optimal technology combination based on the evaluation results of technology integration potential.

9. The enterprise patent database intelligent construction system according to claim 8, characterized in that: The system can dynamically monitor the technological development trends of different industries, regularly update the patent database, and ensure that the analysis results of technological intersections are synchronized with the latest technological developments.

Images