Identifying Faked Hotel Reviews Using Authorship Analysis

Robert Layton, Paul Watters, Oana Ureche · 2013

The use of online review sites has grown significantly, allowing for communities to share information on products or services.These online review sites are marketed as being independent and trustworthy, but have been criticised for not ensuring the integrity of the reviews.One major concern is that of review fraud; where a person (such as a marketer) is paid to write favourable reviews for one product or poor reviews for a competitor.In this research we show a method for determining if two reviews share an author, which can be used to identify if a review is legitimate.Our results indicate a high quality of the method, with an f1-score of over 0.66 in testing data with 40 authors, with most authors having only one or two documents.This type of analysis can be used to investigate cases of potential hotel review fraud.

Read the paper · More papers on PaperTik