A method and system for processing intellectual property data for knowledge clustering

Abstract

The application belongs to the field of information construction, and particularly relates to a kind of intellectual property data processing method and system for knowledge clustering, technical feature recognition and block extraction are carried out on each patent in the knowledge base, and technical classification is carried out, obtaining several knowledge blocks;According to the knowledge block of different patents, the semantic association of different patents is carried out, and the knowledge cluster corresponding to each knowledge block is obtained;User demand data is acquired, data processing is carried out, and technical keywords are converted, technical keywords are matched with knowledge blocks, and knowledge blocks associated with technical keywords are obtained;According to the current knowledge block, all patents corresponding to the corresponding knowledge cluster are found, and after unified processing, knowledge products are obtained;The knowledge product is sent to the corresponding user. Based on the semantic association of knowledge block, the knowledge cluster is formed, the scattered patents are automatically clustered according to the technical logic and functional direction, the problem that the patent quantity is large, the professional is scattered and it is difficult to quickly locate in the field of aviation is solved, and the search and analysis cost of scientific researchers is greatly reduced.

Description

Technical Field

[0001] This application belongs to the field of information technology construction, and specifically relates to a method and system for processing intellectual property data for knowledge clustering. Background Technology

[0002] International aviation giants such as Boeing and Airbus, as well as domestic aviation enterprises and institutions, apply for a large number of patents every year. These patents cover a wide range of new technologies and methods in the aircraft design process. Searching, analyzing, and utilizing this patent information can, on the one hand, help track and grasp the latest technological achievements and development trends in the aviation field, supporting the technological breakthroughs in model development, which has a direct promoting effect on aircraft development. On the other hand, it can accurately grasp the future technological layout and strategic priorities of competitors, helping to establish key research and development directions, which is of great significance for macro-level understanding of the direction of aircraft development work.

[0003] By searching, analyzing, and utilizing patents in the aviation field, we can effectively improve the development capabilities of civil aircraft and the ability to grasp and predict the market, which is of great significance for enhancing our competitive advantage in the civil aircraft market.

[0004] Currently, patent retrieval, analysis, and utilization still rely on free data sources. However, free data sources suffer from drawbacks such as untimely updates, data instability, and incompleteness, hindering the effective conduct of patent retrieval and analysis. Furthermore, many organizations have not established patent information dissemination systems operating within an intranet environment, resulting in a significant amount of valuable technical patent information resources remaining unavailable for use by researchers. This objectively leads to a waste of technical patent resources.

[0005] Therefore, how to deliver patent technology information to R&D personnel in a timely and effective manner is a problem that needs to be solved. Summary of the Invention

[0006] To address the aforementioned issues, this application provides a method and system for processing intellectual property data through knowledge clustering, thereby resolving the problem in the prior art of providing researchers with patent technology information in a timely and effective manner.

[0007] The technical solution of this application is: a commercialized intelligence information support system, comprising:

[0008] Collect a certain amount of design-related patent data and build a knowledge base based on the patent data;

[0009] For each patent in the knowledge base, technical features are identified and extracted in blocks, and then classified into several knowledge blocks.

[0010] Semantically associate different patents according to their knowledge blocks to obtain the knowledge clusters corresponding to each knowledge block;

[0011] After acquiring user demand data and processing it, the data is transformed into technical keywords. These technical keywords are then matched with knowledge blocks to obtain the knowledge blocks associated with the technical keywords.

[0012] Based on the current knowledge block, find all patents corresponding to the knowledge cluster, process them in a unified manner to obtain the knowledge product, and then send the knowledge product to the corresponding user.

[0013] Preferably, a knowledge base is established based on patent data, specifically as follows:

[0014] We collect design-related patent data in the aviation field from multiple national patent databases in real time, and then clean, deduplicate, and index the patent data.

[0015] Patent data is categorized and stored according to a four-level professional tree structure: overall level, system level, subsystem level, and equipment level, to build a knowledge base for patent topics.

[0016] Preferably, technical feature identification and segmentation extraction are performed on each patent in the knowledge base, specifically as follows:

[0017] The process analysis model is used to analyze the process-related technical texts in the patent, down to the specific execution steps and corresponding implementation methods, without further deep-level principle decomposition and parameter refinement;

[0018] A structural analysis model is used to analyze the structural technical text in the patent, extracting the functional structure and functional relationships, without analyzing the specific shape, size, outline and construction details.

[0019] Preferably, the process analysis model includes:

[0020] Based on the BiLSTM-CRF sequence labeling algorithm, process start identifiers, step nodes, and method keywords are identified from the patent text to form step-method pairs. These step-method pairs are used as the smallest units to form process-class knowledge blocks, without further breaking down into sub-steps and underlying principles.

[0021] Preferably, the process analysis model further includes:

[0022] The algorithm employs rule matching and dependency parsing to locate trigger words such as "include, step, adopt, pass, execute", extracts continuous step sequences and corresponding methods, stops at the step layer, and does not enter the parameter layer or principle layer.

[0023] Preferably, the structural analytical model includes:

[0024] Based on the BERT semantic classification algorithm, the function expressions "used for, implemented, completed, provided" in the patent text are identified, and the function name, function object, and function relationship are extracted to form a functional structure unit, without extracting geometric features such as shape, size, angle, and position.

[0025] Preferably, the structural analytical model further includes:

[0026] An entity relationship extraction algorithm is used to extract only the functional roles, connection relationships, and effects of structural components, while filtering out shape features, structural details, and assembly tolerances to form purely function-oriented structural knowledge blocks.

[0027] Preferably, the knowledge blocks are divided into two categories: process knowledge blocks and structure knowledge blocks;

[0028] The process knowledge block only includes: process name, step sequence, method, applicable scenarios, and technical effects;

[0029] The structural knowledge block contains only: structural name, functional definition, functional relationship, target object, and system to which it belongs.

[0030] Preferably, the step of semantically associating knowledge blocks according to different patents to obtain knowledge clusters includes:

[0031] The Sentence-BERT semantic similarity algorithm is used to perform vector encoding and similarity calculation on the steps, logic, methods, and functional structure of knowledge blocks;

[0032] Knowledge blocks with similarity higher than a preset threshold are aggregated into the same knowledge cluster, and the same knowledge cluster corresponds to the same technical direction or the same functional goal.

[0033] Preferably, technical keywords are matched with knowledge blocks to obtain knowledge blocks associated with the technical keywords, specifically as follows:

[0034] The user requirement text is cleaned, segmented, and normalized. Keyword extraction algorithms are used to extract core technical terms and generate a set of Chinese and English bilingual technical keywords.

[0035] Perform fuzzy semantic matching between technical keywords and knowledge blocks, and sort and filter them according to the matching score to obtain knowledge blocks associated with the technical keywords.

[0036] Preferably, after standardization, the resulting knowledge product is as follows:

[0037] Based on the knowledge cluster to which the current knowledge block belongs, all patent information within the cluster is retrieved. The patent information is then processed by standardizing the format, deduplicating, sorting, filtering the legal status, and marking the confidentiality level, generating a patent search list, patent analysis report, or technology early warning information as a knowledge product.

[0038] Another technical solution of this application is: an intellectual property data processing system for knowledge clustering, comprising:

[0039] The knowledge base building module is used to collect a certain amount of design-related patent data and build a knowledge base based on the patent data.

[0040] The knowledge block acquisition module is used to identify and extract the technical features of each patent in the knowledge base, and classify them into several knowledge blocks.

[0041] The knowledge cluster acquisition module is used to semantically associate different patents according to the knowledge blocks of different patents, and obtain the knowledge clusters corresponding to each knowledge block.

[0042] The knowledge association module is used to acquire user demand data, process the data and convert it into technical keywords, and then match the technical keywords with knowledge blocks to obtain the knowledge blocks associated with the technical keywords.

[0043] The knowledge product acquisition module is used to find all the knowledge clusters corresponding to the current knowledge block.

[0044] The intellectual property data processing method and system for knowledge clustering in this application have the following advantages:

[0045] Knowledge clusters are formed based on semantic associations of knowledge blocks, enabling scattered patents to be automatically clustered according to technical logic and functional direction. This solves the problems of large number of patents, professional dispersion, and difficulty in rapid location in the aviation field, and significantly reduces the search and analysis costs for researchers.

[0046] By limiting the process analysis to extract only the steps and methods, and the structural analysis to extract only the functional structure, without delving into the shape, parameters, and underlying principles, we can ensure the integrity of technical features while avoiding redundancy and noise caused by over-analysis, thus significantly improving the standardization and usability of knowledge blocks.

[0047] User needs are standardized and transformed into technical keywords. Through precise matching, semantic matching, and professional tree-level matching, the needs are quickly and accurately associated with patent knowledge blocks without the need for manual screening of each article. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the overall process of this application. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application will be described in more detail below with reference to the accompanying drawings. In the drawings, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The described embodiments are only some, not all, of the embodiments of this application. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application. The embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0050] The first aspect of this application provides a method for processing intellectual property data to perform knowledge clustering, such as... Figure 1 As shown, it includes:

[0051] Step S100: Collect a certain amount of design-related patent data and establish a knowledge base based on the patent data.

[0052] Preferably, a knowledge base is established based on patent data, specifically as follows:

[0053] We collect design-related patent data in the aviation field from multiple national patent databases in real time, and then clean, deduplicate, and index the patent data.

[0054] Patent data is categorized and stored according to a four-level professional tree structure: overall level, system level, subsystem level, and equipment level, to build a knowledge base for patent topics.

[0055] By constructing a thematic knowledge base based on a four-level professional tree, patent data can be accurately matched with the aircraft development professional system, enabling structured storage of patent information, facilitating rapid retrieval, location and management, and enhancing the authority and usability of patent data.

[0056] Furthermore, within different levels of thematic topics, there are also thematic databases, such as the aerospace 3D printing technology database.

[0057] Step S200: For each patent in the knowledge base, technical features are identified and extracted in blocks, and the technology is classified to obtain several knowledge blocks.

[0058] Preferably, technical feature identification and segmentation extraction are performed on each patent in the knowledge base, specifically as follows:

[0059] The process analysis model is used to analyze the process-related technical texts in the patent, down to the specific execution steps and corresponding implementation methods, without further deep-level principle decomposition and parameter refinement;

[0060] A structural analysis model is used to analyze the structural technical text in the patent, extracting the functional structure and functional relationships, without analyzing the specific shape, size, outline and construction details.

[0061] To avoid data redundancy caused by over-parsing, ensure accurate, concise, and controllable knowledge block extraction, and improve the efficiency of subsequent matching and clustering.

[0062] Knowledge blocks are divided into two categories: process knowledge blocks and structure knowledge blocks;

[0063] The process knowledge block only includes: process name, step sequence, methods, applicable scenarios, and technical effects;

[0064] The structural knowledge block contains only: structural name, functional definition, functional relationship, target object, and system to which it belongs.

[0065] Preferably, the process analysis model includes:

[0066] Based on the BiLSTM-CRF sequence annotation algorithm, process start identifiers, step nodes, and method keywords are identified from patent texts to form step-method pairs. These step-method pairs are used as the smallest units to form process-type knowledge blocks, without further breaking down into sub-steps and underlying principles.

[0067] Then, using rule matching and dependency parsing algorithms, the trigger words "include, step, adopt, pass, execute" are located, and the continuous step sequence and corresponding method are extracted. The process stops at the step layer and does not enter the parameter layer or principle layer.

[0068] The BiLSTM-CRF sequence labeling algorithm is used to automatically identify steps and methods, forming the smallest unit of process knowledge blocks, improving the accuracy and consistency of technical feature extraction, reducing manual intervention, and increasing the standardization of parsing.

[0069] Preferably, the structural analytical model includes:

[0070] Based on the BERT semantic classification algorithm, the function expressions "used for, implemented, completed, provided" in the patent text are identified, and the function name, function object, and function relationship are extracted to form a functional structure unit, without extracting geometric features such as shape, size, angle, and position.

[0071] Then, using an entity relationship extraction algorithm, only the functional roles, connection relationships, and effects of structural components are extracted, while shape features, structural details, and assembly tolerances are filtered out to form purely function-oriented structural knowledge blocks.

[0072] Based on the BERT semantic classification algorithm, functional structural units are extracted and geometric features such as shape and size are filtered out, so that knowledge blocks focus on function and role, which is more in line with the needs of technology research and development, patent mining and layout.

[0073] Step S300: Semantically associate different patents according to the knowledge blocks of different patents to obtain the knowledge clusters corresponding to each knowledge block;

[0074] Specifically:

[0075] The Sentence-BERT semantic similarity algorithm is used to perform vector encoding and similarity calculation on the steps, logic, methods, and functional structure of knowledge blocks;

[0076] Knowledge blocks with similarity higher than a preset threshold are aggregated into the same knowledge cluster, and the same knowledge cluster corresponds to the same technical direction or the same functional goal.

[0077] By using Sentence-BERT to achieve semantic clustering of knowledge blocks, patents are automatically aggregated according to technical direction, solving the problem of disorder in massive patents and realizing the modular and systematic organization of technical intelligence.

[0078] Step S400: Obtain user demand data, process the data and convert it into technical keywords, match the technical keywords with knowledge blocks to obtain knowledge blocks associated with the technical keywords.

[0079] Specifically:

[0080] The user requirement text is cleaned, segmented, and normalized. Keyword extraction algorithms are used to extract core technical terms and generate a set of Chinese and English bilingual technical keywords.

[0081] Perform fuzzy semantic matching between technical keywords and knowledge blocks, and sort and filter them according to the matching score to obtain knowledge blocks associated with the technical keywords. This achieves accurate and efficient matching between user needs and knowledge blocks, improving service response speed and accuracy.

[0082] Step S500: Based on the current knowledge block, find all patents corresponding to the knowledge cluster, process them in a unified manner, and obtain the knowledge product; send the knowledge product to the corresponding user.

[0083] Preferably, after standardization, the resulting knowledge product is as follows:

[0084] Based on the knowledge cluster to which the current knowledge block belongs, and by retrieving all patent information within the cluster, the patent information is processed for format unification, deduplication, sorting, legal status filtering, and confidentiality level labeling, generating a patent search list, patent analysis report, or technology early warning information as knowledge products.

[0085] In summary, this application has the following advantages:

[0086] Through a complete process of patent data collection, knowledge block extraction, knowledge cluster association, demand matching, and knowledge product generation and delivery, the system automates and intelligentizes intelligence information support, solves the problems of scattered patent information, low retrieval efficiency, inaccurate demand matching, and non-standard output results, and provides a unified methodological support for the in-depth utilization of patent information in the aviation field.

[0087] By limiting the process analysis to extract only the steps and methods, and the structural analysis to extract only the functional structure, without delving into the shape, parameters, or underlying principles, we can ensure the integrity of technical features while avoiding redundancy and noise caused by over-analysis, thus significantly improving the standardization and usability of knowledge blocks.

[0088] Another technical solution of this application is: an intellectual property data processing system for knowledge clustering, comprising:

[0089] The knowledge base building module is used to collect a certain amount of design-related patent data and build a knowledge base based on the patent data.

[0090] The knowledge block acquisition module is used to identify and extract the technical features of each patent in the knowledge base, and classify them into several knowledge blocks.

[0091] The knowledge cluster acquisition module is used to semantically associate different patents according to the knowledge blocks of different patents, and obtain the knowledge clusters corresponding to each knowledge block.

[0092] The knowledge association module is used to acquire user demand data, process the data and convert it into technical keywords, and then match the technical keywords with knowledge blocks to obtain the knowledge blocks associated with the technical keywords.

[0093] The knowledge product acquisition module is used to find all the knowledge clusters corresponding to the current knowledge block.

[0094] The intellectual property data processing method and system for knowledge clustering in this application have the following advantages:

[0095] Knowledge clusters are formed based on semantic associations of knowledge blocks, enabling scattered patents to be automatically clustered according to technical logic and functional direction. This solves the problems of large number of patents, professional dispersion, and difficulty in rapid location in the aviation field, and significantly reduces the search and analysis costs for researchers.

[0096] By limiting the process analysis to extract only the steps and methods, and the structural analysis to extract only the functional structure, without delving into the shape, parameters, and underlying principles, we can ensure the integrity of technical features while avoiding redundancy and noise caused by over-analysis, thus significantly improving the standardization and usability of knowledge blocks.

[0097] User needs are standardized and transformed into technical keywords. Through precise matching, semantic matching, and professional tree-level matching, the needs are quickly and accurately associated with patent knowledge blocks without the need for manual screening of each article.

[0098] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for processing intellectual property data through knowledge clustering, characterized in that, include: Collect a certain amount of design-related patent data and build a knowledge base based on the patent data; For each patent in the knowledge base, technical features are identified and extracted in blocks, and then classified into several knowledge blocks. Semantically associate different patents according to their knowledge blocks to obtain the knowledge clusters corresponding to each knowledge block; After acquiring user demand data and processing it, the data is transformed into technical keywords. These technical keywords are then matched with knowledge blocks to obtain the knowledge blocks associated with the technical keywords. Based on the current knowledge block, find all patents corresponding to the knowledge cluster, process them in a unified manner to obtain the knowledge product, and then send the knowledge product to the corresponding user.

2. The intellectual property data processing method for knowledge clustering as described in claim 1, characterized in that, A knowledge base is established based on patent data, specifically as follows: We collect design-related patent data in the aviation field from multiple national patent databases in real time, and then clean, deduplicate, and index the patent data. Patent data is categorized and stored according to a four-level professional tree structure: overall level, system level, subsystem level, and equipment level, to build a knowledge base for patent topics.

3. The intellectual property data processing method for knowledge clustering as described in claim 1, characterized in that, For each patent in the knowledge base, technical features are identified and extracted in blocks, specifically as follows: The process analysis model is used to analyze the process-related technical texts in the patent, down to the specific execution steps and corresponding implementation methods, without further deep-level principle decomposition and parameter refinement; A structural analysis model is used to analyze the structural technical text in the patent, extracting the functional structure and functional relationships, without analyzing the specific shape, size, outline and construction details.

4. The intellectual property data processing method for knowledge clustering as described in claim 3, characterized in that, The process analysis model includes: Based on the BiLSTM-CRF sequence labeling algorithm, process start identifiers, step nodes, and method keywords are identified from the patent text to form step-method pairs. These step-method pairs are used as the smallest units to form process-class knowledge blocks, without further breaking down into sub-steps and underlying principles.

5. The intellectual property data processing method for knowledge clustering as described in claim 4, characterized in that, The process analysis model also includes: The algorithm employs rule matching and dependency parsing to locate trigger words such as "include, step, adopt, pass, execute", extracts continuous step sequences and corresponding methods, stops at the step layer, and does not enter the parameter layer or principle layer.

6. The intellectual property data processing method for knowledge clustering as described in claim 3, characterized in that, The structural analytical model includes: Based on the BERT semantic classification algorithm, the function expressions "used for, implemented, completed, provided" in the patent text are identified, and the function name, function object, and function relationship are extracted to form a functional structure unit, without extracting geometric features such as shape, size, angle, and position.

7. The intellectual property data processing method for knowledge clustering as described in claim 6, characterized in that, The structural analytical model also includes: An entity relationship extraction algorithm is used to extract only the functional roles, connection relationships, and effects of structural components, while filtering out shape features, structural details, and assembly tolerances to form purely function-oriented structural knowledge blocks.

8. The intellectual property data processing method for knowledge clustering as described in claim 1, characterized in that, The knowledge blocks are divided into two categories: process knowledge blocks and structure knowledge blocks. The process knowledge block only includes: process name, step sequence, method, applicable scenarios, and technical effects; The structural knowledge block contains only: structural name, functional definition, functional relationship, target object, and system to which it belongs.

9. The intellectual property data processing method for knowledge clustering as described in claim 1, characterized in that, The semantic association based on the knowledge blocks of different patents yields knowledge clusters, including: The Sentence-BERT semantic similarity algorithm is used to perform vector encoding and similarity calculation on the steps, logic, methods, and functional structure of knowledge blocks; Knowledge blocks with similarity higher than a preset threshold are aggregated into the same knowledge cluster, and the same knowledge cluster corresponds to the same technical direction or the same functional goal.

10. The intellectual property data processing method for knowledge clustering as described in claim 8, characterized in that, By matching technical keywords with knowledge blocks, the knowledge blocks associated with the technical keywords are obtained, specifically: The user requirement text is cleaned, segmented, and normalized. Keyword extraction algorithms are used to extract core technical terms and generate a set of Chinese and English bilingual technical keywords. Perform fuzzy semantic matching between technical keywords and knowledge blocks, and sort and filter them according to the matching score to obtain knowledge blocks associated with the technical keywords.

11. The intellectual property data processing method for knowledge clustering as described in claim 8, characterized in that, After standardization, the resulting knowledge product is as follows: Based on the knowledge cluster to which the current knowledge block belongs, all patent information within the cluster is retrieved. The patent information is then processed by standardizing the format, deduplicating, sorting, filtering the legal status, and marking the confidentiality level, generating a patent search list, patent analysis report, or technology early warning information as a knowledge product.

12. An intellectual property data processing system for knowledge clustering, employing the method described in any one of claims 1-11, characterized in that, include: The knowledge base building module is used to collect a certain amount of design-related patent data and build a knowledge base based on the patent data. The knowledge block acquisition module is used to identify and extract the technical features of each patent in the knowledge base, and classify them into several knowledge blocks. The knowledge cluster acquisition module is used to semantically associate different patents according to the knowledge blocks of different patents, and obtain the knowledge clusters corresponding to each knowledge block. The knowledge association module is used to acquire user demand data, process the data and convert it into technical keywords, and then match the technical keywords with knowledge blocks to obtain the knowledge blocks associated with the technical keywords. The knowledge product acquisition module is used to find all patents corresponding to the knowledge cluster based on the current knowledge block, process them in a unified manner to obtain the knowledge product, and then send the knowledge product to the corresponding user.

Images