Document Type

Article

Abstract

This study proposed a system for an automatic question analysis and answering that applies information extraction techniques for extracting the relevant information from questions of the Internet connections. The study uses both ontology, and concept weighting to extract, and classify questions. The information extractions are divided in two steps as follows. First, ontology is used to create a decision tree according to the concepts of questions. Second, concept weighting is applied; the weight of each concept is given by its frequency in the past documents. The results from this study reveal that it is possible to develop an automatic question answering system. The proposed system had been tested with sample questions from the Internet service provider: True Corporation Call Center. Precision of the system is about 0.77 which is good. This study offers useful information regarding the areas of information extraction for Call Center or Web Based Support Systems.

Share

COinS