A study of online review sentiments on rental decision

• Main Authors: Jau-Chiun Wang, 王詔群
• Other Authors: Wei- Jung Shiang
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/06573262720337467258

碩士 === 中原大學 === 工業與系統工程研究所 === 102 === The development of mobile application became more popular and convenient in recent years, and many developers worked on topics related to convenience, localization and high mobility. However, the study of providing individual decision support in the mobile application is not very popular. For example, finding a suitable rental house is different to the past. In terms of decision making, it is hard for a renter to choose the based on the house information posted by the landlord. After signing the contract or moving into the house, the tenant could find some problems, such as water leaking, noise and the attitude of the landlord. This house information could be disclosed by the previous tenant reviews. Therefore, in the mobile application, providing an online review system for renting a house is important. More complete and analyzed reviews can help the tenant to make right decisions. This study proposed to apply the concept of Kansei Engineering to investigate the effect of online reviews on the rental decision making. Typical samples and Kansei adjective words were chosen through the interview and questionnaire survey primarily. Based on differences of word meanings, a weighting model for review samples was built. From the results of correlation analysis, the Kansei adjective review has significant effect on the rental willingness, and the grading results are highly correlated with the rental decision. The adjective words with the most positive effects are ‘‘nice landlord’’ and ‘‘secure and safe’’. On the other hand, the adjective words of the most negative effects are ‘‘bad landlord’’, ‘‘insecure’’ and ‘‘bad internet connection’’. A linear regression model was developed to predict the rental rate of a house with the review grading which was assessed by the weighting model.