An Open-Domain Question Answering System Using Annotated Web Feeds

Abstract

Open Domain Question Answering Systems (ODQA) aim to answer all possible questions, regardless of topic and time. For this to be possible, most current ODQA systems depend on the World Wide Web through search engines, e.g. Google, which provide an abundence of information. Problem arises when search engines require a few days to crawl, archive and index the latest documents depending on the popularity of the website and the search engine’s indexing efficiency. Thus, recent information may not be available as soon as it is published. Our work focuses on capturing and populating current news articles from various trusted resources into a single unified repository. We have implemented a prototype which uses Web 2.0 RSS feed technology to capture the information. This includes an interface which allows other question answering systems to access our repository using various formats, e.g. XML and SQL query. We have also implemented a question answering engine which employs keyword detection, query expansion and rule matching.