Proposal for Soft Computing Applications to Question Answering Systems

Text documents convey a collection of relationships between structured elements, expressed in a natural language discourse. Certain corpora of selected text documents (e.g. a company's product literature and service manuals) contain useful relationships of facts that can be used to answer questions about its content. However, due to the challenges and limitations of natural language understanding by machines, automated systems that can deduce such relationships and answer queries about them have not been developed to an acceptable quality.
The proposed project aims to advance the state-of-the-art of information retrieval, by combining search, deduction, semantic modeling, world and domain knowledge, and advanced soft computing paradigms (computing with words) to research and develop question answering systems. The objective is not just to retrieve the most relevant document or text passage from the corpora via keyword matching, but to (i) better "understand" the question and its intent, and (ii) synthesize a response from known facts and relations that are relevant to the question.