Knowledge management techniques towards answering engine

Abstract

The growth of Internet era has brought magnitude of information at the disposal of end user being mobilized with the speed of electron. But this bless of information comes with a curse of ‘information overload’. Answer engine has come out as a well-accepted solution to this problem as it target to directly provide answer to the user’s question. At present, several hurdles exist in the path of a robust answer engine capable of handling various types of complex questions and they have been mostly known as inefficient in handling natural language text. This thesis presents a study, implementation and some contribution to various aspects of this question answering approach. Notably a novel approach of summarization has been tried that works on the principle of capturing relation between salient concepts present in text. The work also presents a Feature Drift model for tracking concept drift, which is a key ingredient in construction of user model required for personalization at various stages. Finally, a representative implementation of a factoid Answer Engine and a Search Engine built for the domain of DA-IICT has been described.