Design And Implementation Of Patent Information Search System Based On Semantic Matching

With the advent of the era of artificial intelligence,the output of a large number of patents and scientific research achievements has gradually accelerated the pace of technology investment in the industrial field.However,the following problem is the distinction and adaptation of professional domain knowledge involving all walks of life.Due to the large number of relevant patent information and scientific research results,the text in specific fields needs to be extracted to meet the detailed technical needs of industry and enterprises.This situation depends on the detailed analysis of experts in specific fields,and the efficiency is low.Aiming at the actual needs of enterprises and the problem of patent matching,this paper designs a set of patent keyword retrieval and demand matching system to manage and solve the matching service of specific scientific research achievements.The system adopts a self-trained deep learning algorithm to improve the accuracy of text matching and search efficiency.The whole system mainly includes the following two main designs:(1)Matching algorithm model between requirements document content and patent and scientific research achievement databaseThe data used to build the algorithm model in this paper is the industry specific demand text data and scientific research achievement text data set provided by the Ministry of science and technology.The construction process adopts the bottom-up model,combined with the relationship categories pre identified by experts in various fields,to realize the construction of the overall algorithm model.(2)Design and implement search knowledge service systemThe whole system is divided into four modules:system management,full-text retrieval,core matching module and transaction module.The system management module solves the management problems of user information,patent and scientific research achievement information and field category information.The full-text retrieval module solves the user’s demand for short term search.The purpose is to give priority to full-text search in the case of short words,that is,the distributed full-text search function provided by elasticsearch.The main task of this module is to meet the search needs of users for simple short words.The main function of the core requirements matching module is to realize the analysis and reasoning of the text content and quickly match the most suitable items in the scientific research achievements database according to the requirements documents,requirements information or long text submitted by users.The transaction module mainly includes the acquisition of the full text of the patent,which is used by business personnel to purchase patents or achievements that meet the needs of the enterprise.