Abstract
Scholars increasingly seek to investigate differences between authentic and manipulative online reviews. A common line of research argues that authentic and manipulative reviews are distinguishable based on three textual characteristics, namely, comprehensibility, informativeness and writing style. Although recent studies have analyzed differences between authentic and manipulative reviews in terms of these textual characteristics, they often lack in terms of methodological rigor. For one, datasets used for analysis are not always representative. Moreover, only few machine learning algorithms are used to classify authentic and manipulative reviews. Recognizing the value of methodological rigor, this paper extends prior studies by examining textual differences between authentic and manipulative reviews using a more representative dataset. Moreover, authentic and manipulative reviews were classified using a voting among multiple classifiers that had been used in recent literature. The implications of the results are discussed.
Original language | English |
---|---|
Pages | 77-83 |
Number of pages | 7 |
Publication status | Published - 2015 |
Event | Science and Information Conference - London, United Kingdom Duration: 28 Jul 2015 → 30 Jul 2015 http://saiconference.com/Conferences/SAIConference2015 |
Conference
Conference | Science and Information Conference |
---|---|
Country/Territory | United Kingdom |
City | London |
Period | 28/07/15 → 30/07/15 |
Internet address |