어떤 리뷰가 도움이 되는가? 영화 장르에 따라 달라지는 속성기반 리뷰의 유용성

초록

This study empirically examined how textual attributes embedded in IMDb movie reviewsinfluenced helpfulness evaluations and how these effects varied across movie genres. A total of29,797 reviews from 33 movies posted on IMDb were analyzed by quantifying six major textualattributes (character, acting, directing, storytelling, visual effects, and music) and applyingthe Heckman two-stage selection model to correct for voting selection bias. The analysis revealed a U-shaped relationship between review length and the likelihood ofbeing voted as helpful: shorter reviews were evaluated as less helpful, whereas longer reviewstended to receive higher helpfulness votes once they exceeded a certain length. In contrast, thetotal amount of attribute-related information exhibited a monotonically decreasing relationshipwith helpfulness, indicating that reviews containing more attribute-related information wereless likely to be evaluated as helpful. The proportion of irrelevant information also negativelyaffected helpfulness. Furthermore, the effects of textual attributes on helpfulness differedacross attributes, and even the same attribute exerted different effects depending on the moviegenre. For example, information on visual effects had a positive impact in genres such asanimation, where visual quality is highly valued, whereas information related to charactersand narrative structure had a stronger impact in action and thriller genres. These findings demonstrated that helpfulness evaluations were determined not by the sheeramount of information but by the contextual fit between review content and consumerexpectations. This study quantitatively validated information fit theory and further explicated how electronic word-of-mouth (eWOM) effects differed across genre contexts. The results providepractical implications for movie studios and review platforms in designing genre-specificstrategies for review generation and recommendation.

키워드