The Study of Social Network Opinion Mining in Hakka Tourism Industry – In Case of Hsinchu and Miaoli Sightseeing Spots

• Main Authors: LAN, CHI-WEN, 藍啟文
• Other Authors: GUO, JIUNN-LIANG
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/e4c88s

碩士 === 國防大學 === 運籌管理學系 === 106 === Hakka Affairs council announced the Romantic Provincial Highway 3 (Taiwan) campaign to promote the cultural value-added of Hakka village in tourism industry. Using Hakka culture tourism destinations from Beipu and Nanzhuang of Taiwan as examples, we conducted the text mining technique to identify the perceptions and attitudes of tourism experience from social media - Google maps reviews. This research aims to conduct classification methods together with the TF-IDF weighting schemes to uncover salient terms which can best describe the classification model of social sentiment comments. Accordingly, the designed framework adopts seven classification algorithms with 10-folded cross validation process to determine the optimum model. As a result, the sentiment classification system suggests that Maximum Entropy and Bagging outperform others in accuracy of evaluation metrics. Based on the collecting and analyzing tourist experiences from both selected tourism destination, the corresponding tourism values and overall perception can be observed from the perspectives of tourists. According to the findings, there are four negative opinions, unclear tourism positioning, traffic problems, unhealthy dietary needs, and lift-up prices which can provide vital information to related tourism bureau in decision-making of future policy on tourists’ needs and attitudes.