Research And Implementation Of Question Answering System Based On Traditional Chinese Medicine Semantic Web

People want to get the answer to the question in the first time. For this reason, Question Answering has always been a promising research area of artificial intelligence and natural language processing. With the support of hardware as well as the development of internet and AI technology, excellent QA systems such as IBM Watson were built. But most of them are open-domain QA, the data of which is not reliable, and the questions it answers are always elaborately constructed. As a consequence, building a QA in specific-domain has more realistic significance.As a treasure of traditional culture, Chinese herbal medicine contributes a lot to the breeding of the Chinese nation, and interests more and more people. The Nobel Prize for Artemisinin makes it a great upsurge in learning and studying Chinese herbal medicine at present. This paper is based on huge amounts of data in specific domain, and focuses in research and implementation of the QA system for Chinese herbal medicine, which has a significant meaning.Our work is summarized as follows:1) We study the theory of semantic web first, then design the top level ontology in traditional Chinese medicine. After reconstructing the database, we implemented a tool which can be used to extract information to build semantic web automatically. Finally we obtained about 3 million triples.2) After research on QA, we propose an approach of parsing natural language to SPARQL, which is based on patterning matching, domain knowledge and machine learning. By combining different features as an input, the results of the experiments show this approach works well.3) Based on the research above, we design and implement the QA based on TCM semantic web. The test of questions in reality shows that the methodology is workable in specific domain QA and the system we developed is efficient and practical.