Open domain factoid question answering system

Mankind has recently produced a vast amount of data. According to IBM, every day we create 2.5 quintillion bytes of data. To find the relevant information in this huge body of data, we use search engines. These search engines locate the relevant data according to a user's query and return a list of results to the user, after which the user finds what he is seeking in this list. However, users may not wish to read a list of results and may prefer short, relevant answers to their query. For this reason, there is much research on Question Answering Systems. The task of these systems is the acquisition of relevant information to questions asked by users. In this thesis, we developed a question answering system called Prime and describe how it works. Prime is a factoid question answering system. The term factoid is defined as "an item of unreliable information that is reported and repeated so often that it becomes accepted as fact." Therefore, it gives only an answer to a user's factoid question. We also describe how we extract question triples (subject-predicate-object) based on dependency graphs and build SPARQL queries to query DBpedia v based on extracted triples in addition to describing how we map entities in questions to DBpedia entities. We evaluated our system with QALD-5 Question answering challenge questions and show that our system promises worthy results. In the QALD-5 training test questions after excluding any unnecessary questions, our system answered 43% of the questions. In the test questions, our system answered 41% of the questions. Furthermore, we describe the evaluation results and evaluation metrics and compared our system with other QALD-5 participant systems.