Sunday, 20 May 2012

QandA: An open-domain question answering system

QandA is my graduation project. It is a web-based natural language open-domain question answering system. Its goal is to retrieve best relevant short answers to users' English questions that are written in natural language, rather than retrieving complete documents or best matching passages like what most of the information retrieval systems do. It extracts answers from any collection of texts, by combining syntactic and semantic analysis, and using different knowledge sources.

Here is a demo that shows the QandA system operating on a supplied passage, and it answers the proposed questions in real time.