Long Text Topic Mining and Clustering Analysis Base on Doc2Vec-LDA and K-Means

In this paper, the research data set takes high-level journal papers in the field of Agriculture and Fishery as the research object, and divides the text data set into two data sets for data cleaning and prepossessing according to the theme. Firstly, the academic paper abstract corpus is trained by splicing Doc2Vec text vector and LDA topic feature analysis model, and the mixed text feature vector matrix is obtained. LDA is used to deeply mine and analyze the topic hot words in the fields of Agriculture and Fishery. Finally, K-Means algorithm is used to cluster and visualize the classified feature word vectors.