Researchers at Cornell have developed a software which is able to distinguish fake reviews from real ones. The software examined around 800 reviews of hotels in Chicago and fake reviews were picked up with 90% accuracy.
According to Myle Ott, a graduate student, though the study has been done for the first time, the software will help to do away with fake reviews.
Manual sorting of fake reviews mixed with original ones did not bear much fruit. The judges who were selected to do the exercise, failed to distinguish the fake from the original, which the software could do with far greater accuracy.
A subtle way of analyzing the terms in the text, was the method adopted by the software, in the sorting. Some concrete words are used in original reviews, which the fake reviews do not take into account. Also fake accounts use more of verbs whereas the real ones use more of nouns.