Finding Answer in Biomedical Documents for Factoid Queries by Using Statistical Approaches: Array

Information Retrieval System as well as question answering is task of extracting short and precise answer that relevant to a specific question rather than giving a long-ranked list of documents. Due to the exponentially growing number of biomedical documents and literatures, the information seekers such as biomedical expert or medical specialist faces many challenges to get access the data. The current search engine only provide user ranked list of biomedical documents rather than directly offer the short answer or information to the user. For biomedical question answering, most of the researchers used semantic-based approach which means it is dependent on the complicated additional resources such as UMLS and MetaMap. Hence, the aim of this research is to solve these problems by using statistical approaches, which are Boolean Model, Vector Space Model and OkapiBM25 Model to obtain the information from relevant biomedical documents as well as to answer the biomedical factoid question. Factoid queries are the questions that will providing concise fact in others word is short answer. The results using these approaches are then evaluated using manual and ROUGE metrics that included measurement of precision, recall and F-measure. The outcomes shown that Vector Space Model is outperformed in extracted the long document as answer than other models. While OkapiBM25 Model has the best performance in retrieved short documents. The detail analysis of research results is being conducted and the possible future works are suggested.