Medical Record Semantic Analysis Based on Weighted LDA

Latent Dirichlet allocation (LDA) and other modified topic models have become the prevalent tools for semantic analysis and text data mining. With the rapid development of the medical information, large amount of data has been accumulated in the form of text, while most of which recording in a confused structure. Based on LDA, this paper proposes a revised approach, which enlightens an idea of weighting. This paper also demonstrates the desirable performance of the modified method, thus yields the proof of its effectiveness.