sammy Posted March 27, 2010 at 05:27 AM Share Posted March 27, 2010 at 05:27 AM For ratings that are subjective, the listings owner tends to exaggerate the ratings. To solve this problem an index to measure the difference between what the listing owner rated and what the reviewers rated. Statisticians do not only use mean/median for analytics, but also use standard-deviation. To improve the quality of the ratings it is best that jReviews also implment standard-deviation. For example: Pizza Hut Reviews Median (total:2) - Median instead of Mean ignores outliers Overall: 3.75 Service: 4/5 Ambiance: 4/5 Value: 5/5 Food: 2/5 Reviews-1 By User-1 Overall: 3.25 Service: 3/5 Ambiance: 4/5 Value: 5/5 Food: 1/5 Mismatch Index of Review-1 = Service (1) + Ambiance (0) + Value (0) + Food (1) = 2 Reviews-2 By User-2 Overall: 3.75 Service: 5/5 Ambiance: 4/5 Value: 3/5 Food: 3/5 Mismatch Index of Review-2 = Service (1) + Ambiance (0) + Value (2) + Food (1) = 4 > The Mismatch Index of each review will be showing next to each review. > The Mismatch Index of each user will be added up for each of his reviews at user-level > The Mismatch Index of each user can be compared against the mean of other users. I think this will improve the quality of the ratings and weave out dubious reviews. Link to comment
Triplex79 Posted September 21, 2012 at 10:10 AM Share Posted September 21, 2012 at 10:10 AM This section is only visible with a valid subscription. If you have a valid subscription, please login. Link to comment
Recommended Posts