Developing a Platform That Supplies Processed Information from Internet Resources and Services

In this paper, a Turkish question answering system that extracts most suitable answers from Internet services and resources is described. During the question-analyzing period of our question answering system, the question class is determined from lexical-morphological properties of the words appearing in the question. In addition, some other properties of the question are predicted and various solution methods are tried in order to get the right answer. Furthermore, to increase the success rate of the system, Wordnet platform is used to find synonym words and other related words of question words. In the information retrieval phase of the system, the system utilizes Tim Berner Lee´s semantic web approach, instead of classic search engine approach. Using extracted features of given questions, subject-predicate-object triples are fetched from DBpedia, which extracts structural information from Wikipedia articles. From obtained triples, the final answer is formulated. For searching and getting Turkish equivalent of the information, Wikipedia Search API and Bing Translate API are used.