Automatic Question Answering System Based on Contextual Interaction
In the question-answer system aimed at course learning, the serial questions asked by user should not be processed respectively, but need to be replied according to context, and the premise is to examine the relevance between user’s current question and quondam questions. An automatic question-answer system is designed which can carry out contextual interaction. The relevance of user’s questions is detected with SVM, the elliptical sentences in the following questions are recovered, and final answers are extracted from ontology-knowledge-base after classifying questions and retrieving information. The experiment result indicates that the F measure of question detection reaches 88% and the accuracy rate of questions is 69.5%.
H. X. Liu et al., "Automatic Question Answering System Based on Contextual Interaction", Applied Mechanics and Materials, Vols. 58-60, pp. 1219-1225, 2011