Do reviewers’ words affect predicting their helpfulness ratings? Locating helpful reviewers by linguistics styles

Locating helpful reviewers in opinion-sharing communities is an important issue. Numerous studies that examine this using social relations have some shortcomings. This study investigates language use, differing from person to person, and develops a novel prediction model to alleviate the problems. We identify four stylistic aspects and explore their impacts on predicting reviewers’ helpfulness ratings. The analyses show that the proposed model can more accurately locate helpful reviewers than the baseline model. In addition, reviewers’ words impact more than social relations do, although a combination of these will boost prediction performance to a greater extent than one alone.