Measurement of Word Similarity for Diabetes Question Answering System

Abstract

Diabetes is a chronic disease that cannot be cured and is a major problem for the public health of Thailand. The Department of Disease Control predicts that by 2025 there will be more than 7.41 million people in Thailand with diabetes. Continuous self-care for people with diabetes is one method that helps to reduce the incidences of complications arising in already compromised body systems affecting the lives of patients. This research, therefore, presents a measure of the similarity of words in Thai question-answering systems for diabetes by using Cosine, Dice and Jaccard methods to compare the effectiveness of finding answers for the benefit of people who want to know about the initial symptoms of diabetes and self-care for people with diabetes. The preliminary results from the study comparing answer finding efficiency using the question-answer similarity measurement methods found that Cosine was the most effective in finding answers with a precision value of 92.50%, followed by Jaccard and Dice which had precision values of 80.28% and 52.50% respectively.

Keywords

• similarity
• diabetes
• question answering system