Investigating reviewers' intentions to post fake vs. authentic reviews based on behavioral linguistic features

Growing interest in peer-generated online reviews for product promotion has incentivized online review manipulation. The latter is challenging to be detected. In this study, to discern reviews that are likely authentic vs. fake, we leverage interpersonal deception theory (IDT) and then investigate verbal and nonverbal features that reflect reviewers' intentions to post fake vs. authentic reviews by using topic modeling techniques. Our findings show topic differences between fake vs. authentic reviews. Based on the results, review manipulators tend to write reviews recommending particular movies, while authentic reviewers are likely to provide movie content information in their reviews. Also, we reveal that review manipulation happens at the early stage of product diffusion and contributes to increasing review ratings. Lastly, we discover that manipulated/fake reviews are more informative and positive. Our findings contribute to extend research on online fake reviews literature by innovatively examining review-writing intentions with topic differences, sentiment, and informativeness. To the best of our knowledge, this is the first attempt to introduce topic factors in the fake review detection literature.