Abstract: Latent Semantic Analysis (LSA) is a mathematical model that is used to capture the semantic structure of documents based on word correlations in them. In-spite of being completely independent of any external sources of semantics, LSA captures the semantic structure quite well. However, previous work in the literature show that including any supplementary information in LSA influences the model's ability to capture the semantic structure of documents. The work presented in this paper is to investigate how supplementary information influences the semantic structure of documents.

Keywords: Dimensionality Reduction, LSA, Semantic Structure, Supplementary Information, SVD.