A TEXT CLASSIFICATION METHOD BASED ON LATENT TOPICS
Yanshan Wang, In-Chan Choi · 2012
Latent Dirichlet Allocation (LDA) is a generative model, which exhibits superiority over other topic modelling algorithms on latent topics of text data. Indexing by LDA is a new method in the context of LDA to provide a new definition of document probability vectors that can be applied as feature vectors. In this paper, we propose a joint process of text classification that combines DBSCAN, indexing with LDA and Support Vector Machine (SVM). DBSCAN algorithm is applied as a pre-processing for LDA to determine the number of topics, and then LDA document indexing features are employed for text classifier SVM.