The Role of Semantic Information in Learning Question Classifiers

Question Classification is commonly used in question answering systems to perform a semantic classification of the target answer in an effort to provide additional information to downstream processes. It is different from the common text categorization task in the sense that questions are relatively short and contain less word-based information compared with classification of the entire text. This work presents a machine learning approach to this task. Our approach is to augment the questions with syntactic and semantic analysis, as well as external semantic knowledge, as input to the text classifier. It is shown that, in the context of question classification, augmenting the input of the classifier with appropriate semantic category information results in significant improvements to classification accuracy.