Japanese named entity recognition for question answering system

Abstract

Current question answering (QA) systems usually contain named entity recognizer (NER) as a core component. NER is an important and difficult task in computational linguistics. It plays an important role in natural language processing application such as Question Answering, Machine Translation, and Information Retrieval etc. NER includes the identification and classification of certain proper nouns (like location, organization, person, data, money and others) in a text. The purpose of our study is to recognize and extract the exact Japanese sight seeing domain named entities. It is a basic step for the following processing: question analysis and keyword extraction information retrieval. As well as, through doing the named entity recognition, we consider that it can mine exact information from text document to respond to user. This paper describes how to do the Japanese sightseeing named entity recognition due to we are constructing a Japanese sightseeing question answering system. We adopt the hybrid method which combined with machine learning and rule-base method. In the experiment of Japanese sightseeing domain named entity recognition we have got excellent precision and recalling rates. It shows that our method is effective and can be used in a practical question answering system.