Research And Application Of Patent Map Service System

The innovation of science and technology is a hot frontier field in the development of the world today.With the gradual maturity of the forms of innovation systems in various countries,the rise of regional innovation systems has received national attention.In the development process of regional innovation,the most critical driving force is introduction of talents and experts.The most intuitive carrier for the progress of science and technology and authoritative information about experts is patent documents.In-depth analysis of patent documents can obtain more comprehensive expert information.Patent map is a more intuitive and reliable tool for mining patent information resources.It can extract and classify patent text data with complex content organization,and present it to users in the form of data visualization charts.This article focuses on regional heat maps,inventor-unit relationship maps,patent word clouds,and patent clustering scatter diagrams combined with a functional patent map system to provide reliable talent information and technological development in response to the needs of decision makers in scientific and technological development.In this way,the role of scientific and technological innovation as an aid to decision-making can be realized.The main research content and experimental results of this article are:The content structure of the patent text is analyzed,and the fields required for patent clustering are selected.The method of obtaining patent data was studied,and a resource pool related to patent data was established.A feature engineering method based on the patent text of ALBERT is proposed.The patent text is calculated through the ALBERT model to calculate the text vector to provide data support for the subsequent clustering algorithm.Input the patent text feature vector into the K-means clustering algorithm to perform cluster analysis on the patent text,and compare it with traditional feature engineering methods,such as word2vec,vsm,etc.Experiments show that the feature engineering method selected in this paper has a better effect on patent document clustering.Based on the results of experimental clustering,visualize the clustering data in the system,and realize the system's regional heat map,inventor-unit relationship map,patent word cloud,and patent cluster scatter points The function of map combination finally uses Echart to make the patent map.This paper designs and produces a patent map system that can perform in-depth data mining of patent texts,and presents intuitive data in the form of inventor-institutional unit,regional heat map,and patent word cloud,which can better meet the needs of users for patent analysis.The feature engineering method used in this system is more suitable for the clustering of patent documents and has certain practicability.