A Latent Semantic Analysis-Based Approach to Geographic Feature Categorization from Text

Geographic feature categorization from text addresses the need for querying and finding geographic features from text documents. Although many text classification techniques have been developed, there are limitations to apply to geographic features due to the uniqueness of the geography features. In this paper we propose a method to classify geographic features based on latent semantic analysis and domain knowledge. The empirical experiment indicates that the proposed method achieves satisfactory categorizing effectiveness.