Design And Implementation Of Patent Recommendation System Based On Knowledge Graph

At present,China is actively promoting the construction of an innovative country,advocating innovation and entrepreneurship among the masses,encouraging the innovative development of domestic enterprises and supporting innovation among all people.Users need to acquire a large amount of scientific and technological knowledge related to its research,grasp the scientific research dynamics and hotspots,discover breakthroughs in technological innovation,and promote technological progress.As an important knowledge carrier of scientific and technological innovations,patents contain a large amount of information on scientific and technological knowledge and technological development,which is one of the important ways for enterprises and innovators to obtain relevant information and knowledge.In order to explore the knowledge and associations contained in the patent data,and facilitate users to quickly acquire these knowledge and relationships,this paper designs and develops a patent recommendation system based on knowledge graph,it recommends related patents for requirements,supports users' innovation work,and maintains core for enterprise users.Competitiveness,innovative development,and an important guiding role for innovators to understand the status quo of scientific research and discover innovation points.The system developed in this paper uses visualization technology to visualize the evolution history of patent technology,hot research areas and the relationship between knowledge,and to recommend related patents for innovators in order to support users'innovation work.The background function module mainly includes multi-source data acquisition,patent data processing and patent data knowledge graph construction.The functional modules provided by the system mainly include patent knowledge retrieval,patent recommendation for invention requirements and patent multidimensional display.The multi-source data acquisition module mainly implements functions such as crawling and storing patent data,encyclopedia knowledge data and technology blog data.The patent data processing module includes functions such as data cleaning,acquiring word vectors,subject classification,and extracting keywords.The patent data knowledge graph building module includes functions such as entity extraction,relationship and attribute extraction,entity recognition,construction of triple data and knowledge data storage.The patent knowledge retrieval module mainly realizes the search function based on knowledge graph,including four types of search:patent,author,applicant and subject word,and realizes the visualization of search results by means of force-oriented map,histogram and dynamic vocabulary.The patent recommendation module for:invention needs mainly recommends relevant words,patents and core sentences for the user's needs,and helps users to develop and solve the initial solution.The patent multi-dimensional display module mainly includes functions such as automatic labeling of patent content,keyword extraction and statistics,and analysis in the patent field.In the process of system development,this paper proposes a Keras-based word segmentation model to improve the word segmentation effect of professional nouns in patent data,and improves the TextRank algorithm to extract patent data keywords.The K-means-based entity recognition algorithm model is proposed for the recognition and differentiation of entities in the knowledge graph construction,and a multi-dimensional scoring algorithm model is proposed for patent ranking recommendation.The system adopts B/S architecture for system implementation,uses Neo4j graph database to store patent data knowledge graph,and D3.is technology to realize data visualization.This paper focuses on the design and implementation of a patent recommendation system based on knowledge graph.At present,the system has more than 2 million patent data,more than 8,000 encyclopedic knowledge and more than 20,000 technology blogs,which are used to support the operation of the system.The system is now in the demonstration application stage,and has helped more than 10 demonstration application companies to develop more than 60 products and technological innovation programs,and achieved good application results.