Research of Topics Discovery and Tech Evolution Based on Text Preprocessed Latent Dirichlet Allocation Model: Research Topic Analysis in GaN Tech Field

摘要

Computational Science and Data Science are inspiring the intelligent analysis and information service today. Machine learning text analysis is changing the traditional analysis methods. This article discusses the benefits of unsupervised learning approaches in patent text mining. Patent data of GaN industry were preprocessed by filter model based on NLTK Toolkit to identify the tech terms and then clustered them based on Latent Dirichlet Allocation model to find the latent topics which were visualized. Based on group operation, new emerging terms ranked by TFIDF through every year were used to reveal the research and development focused evolution. This research offers a demonstration of the proposed method based on 26,854 GaN patents. The results show 20 Research and Development topics with tech terms in GaN industry and present a Research and Development focus evolution based on new emerging terms every year, which provides a clue for more detaied analyses later. Our results show an efficent way to find technology focused evolution from a large scale text data.