Biomedical Question Types Classification using Syntactic and Rule based Approach

A Machine Learning-based Method for Question Type Classification in Biomedical Question Answering

“…Participants [19,34,20] of the challenge however may not have taken into account the problem of biomedical question type classification. In our previous work [21], we have proposed a set of patterns which could classify biomedical questions into yes/no, factoid, and summary categories. However, we have not dealt with all types of questions such as list questions and it is difficult to distinguish between factoid and summary questions.…”

“…With a view to developing an automatic biomedical QA system, the types of biomedical questions should be identified automatically by the system. For that reason, we have previously provided a set of patterns [21] to classify biomedical questions into one of the three categories: yes/ no, factoid and summary. However, this study does not deal with all types of questions.…”

A Generic Document Retrieval Framework Based on UMLS Similarity for Biomedical Question Answering System

“…Finally, the type of question identified in the question processing and the candidates answers generated in the document processing are used in the Answer Processing phase in order to extract the final answer. Previously, we have addressed the problem of biomedical question classification [16] and in this paper we are interested on biomedical document retrieval, which is an important component of biomedical QA systems. As we mentioned earlier, the task of biomedical document retrieval is to find a list of relevant documents that are likely to contain the answer.…”

“…Figure 1 illustrates the generic architecture of a biomedical QA system. For a given biomedical question written in natural language, the Question Processing phase aims to analyze the question and create IR query, identifying the type of question as well [16] . Indeed, the first task is called Query Reformation and the second is called Question Classification [11,16].…”

“…For a given biomedical question written in natural language, the Question Processing phase aims to analyze the question and create IR query, identifying the type of question as well [16] . Indeed, the first task is called Query Reformation and the second is called Question Classification [11,16]. The document Processing phase allows to process the returned documents by an IR system and provides candidate passages which could probably contain the answer.…”