Semantic question answering (SQA) was the research study regarding the natural language processing. The purposes of this study were 1) to encourage the users to query via the computer with the semantic natural language, and 2) to obtain the concise, accurate, and relevant to the users? needs. Recently, it was found that the research studies have encountered the problems with semantic communication, flexibility, and accuracy of processes, especially the process of question classification. It was considered the vital process of developing the semantic question answering system. Thus, this paper attempted to propose a semantic question classification for the question answering system using the linked open data approach. It proposed the problem-solving technique for question classification through semantic grammar rules derived from the questions based on principles of English grammar. Moreover, the linked open data, WordNet and DBpedia were implemented to solve the problems of words similarity and question classification through the question classification taxonomy consisting of six main classes and fifty subclasses as standards for question classification. Besides this, the dataset from the question sets of TREC with one thousand questions were also implemented. The evaluation indicated a high accuracy of question classification with the total scores of precision, recall, and F-measure, at 92.82%, 95.16%, and 93.97%, respectively.
展开▼
