Non-Factoid Japanese Question Answering through Passage Retrieval that Is Weighted Based on Types of Answers

We constructed a system for answering nonfactoid Japanese questions. We used various methods of passage retrieval for the system. We extracted paragraphs based on terms from an input question and output them as the preferred answers. We classified the non-factoid questions into six categories. We used a particular method for each category. For example, we increased the scores of paragraphs including the word “reason” for questions including the word “why.” We participated at NTCIR-6 QAC-4, where our system obtained the most correct answers out of all the eight participating teams. The rate of accuracy was 0.77, which indicates that our methods were effective.