The Model Research of FAQ Answering System Based on Concept

Abstract

At present, the FAQ (Frequently-Asked Question) answering system cannot understand the user's questions at the concept level, so its efficiency needs to be improved. In this paper, a model of FAQ answering system based on concept is proposed. The system model consists of three components, which are concept-based preprocessing, extraction of question-answer pairs, and concept-based matching of sentences. On the one hand, the model proposes an index-based extracting method of question-answering pairs, to combine the extraction with the index mechanism for improving the speed of extraction. On the other hand, the model expands the user's question at the concept level, and proposes a concept based matching method between sentences, to match the user's question with the question-answering pairs. The concept based matching method concerns the synonymous meanings between sentences at the concept level. In addition, the experimental result shows the matching method is an efficient method.