A Hybrid Approach for Natural Language Querying Segmentation for Tourism Ontology

Abstract

We propose an approach to the interpretation of natural language queries, particularly relevant to the tourist industry in Thailand, by applying natural language queries against a tourism ontology containing tourism information specific to Thailand, and in the Thai language. Queries in Thai language are difficult to segment into words and meaningful phrases given that Thai has no word separation, such as in European languages which have a space between words, meaning specific Thai natural language processing is required. This paper demonstrates the identification and comparison of various methodologies currently available for segmenting natural language phrases, which allowed us to develop a hybrid approach based on aspects of natural language processing drawn from the methods analyzed. Our primary contribution is the hybrid approach, which was applied particularly to the queries and questions likely to be posed by Thai tourists in Thai language. Our discussion presents and describes the framework of the proposed methodology, which applies Thai Word Segmentation together with Trigram and the Name Entity method, together with our evaluation of the query response accuracy achieved, which collectively was 99%. We are confident that the proposed approach can be applied for developing any semantic searching application that allows natural language query, otherwise than for tourism in Thailand, our immediate concern.