Research On Product Patent Design Knowledge Extraction Technology Based On Text Mining

Early patent mining was based on structured entry data.In recent years,with the rise of big data technology and machine learning technology,patent mining has gradually developed toward the deep mining of unstructured data as a data source.First of all,this paper reviews the current status of patent term extraction,document modeling,patent classification clustering,and knowledge extraction for deep patent exploration.Based on the existing research,this paper puts forward the main research contents of this paper,the feature word selection and document modeling research based on semantic information,the second-class cluster analysis of patent literature,and the research of design knowledge based on clustering results.Secondly,it introduces the text mining technology used in the research process,and summarizes the technical characteristics and general flow of text mining.This paper combines the word part of speech to improve the patent keyword recognition algorithm,and proposes a method of selecting the patent document feature words by combining semantic information.The TF-IDF value is calculated for the selected feature words,and the vector space model(VSM)is established to complete the mapping from natural language to mathematical model.Based on the idea of combinatorial innovation,an algorithm for extracting design knowledge from the clustering results of two types of patents is proposed.In this algorithm,VSM is taken as the research object,and the cosine similarity is used as the standard for measuring text similarity.The k-means algorithm is used to cluster the data set,and the novel values of design knowledge are calculated and sorted in the same clustering result.The compatibility between reference design knowledge between clusters forms a design knowledge matrix for designers to refer to.Designers can analyze matrix elements one by one,and select knowledge with innovative design points to improve the design.Finally,combining the algorithms and research results proposed in the research,the theory is applied to practice,and the product patent design knowledge extraction system is designed and implemented.The product design knowledge extraction can be carried out for patent documents.