基于LDA主题模型的专利文本分析应用研究

为了解决运用传统的专利文本分析方法中出现的分类粗糙、科学性不强等问题,提出使用基于Latent Dirichlet Allocation(LDA)主题模型的专利文本分析方法。以我国转基因玉米育种领域专利文本为研究对象,对专利文本主题进行划分。通过实验表明LDA主题模型对专利文本的划分和热门领域的分类准确、有效。并得出结论:对专利文本使用LDA主题模型建模可以实现对专利文本的分类,可以实现对热门领域主题分类、子领域细分,并对未来发展态势给出意见。

Research on Application of Patent Text Analysis Based on LDA Topic Model

In order to solve the problems of rough classification and scientificity in the traditional patent text analysis method, a patent text analysis method based on the Latent Dirichlet Allocation (LDA) theme model is proposed. Taking the patent text of the field of genetically modified maize breeding in China as the research object, and divide the subject of patent text. Experiments show that the classification of patent texts and the classification of hot areas are accurate and effective using the LDA topic model. Using the LDA topic model to analyze the patent text can realize the classification of patent texts, and can realize the division of sub-topics and sub-field subdivision, and realize the hot competition situation analysis under each theme. Give opinions on the future development trend.