i'm implement a system using sentiment analysis which will help to users to get a real user experience about the product.. using sentiment analysis i recognize no of positive comments and no of negative comments and sentiment score of each product. now i want to rate the product using those factors.. is there any algorithm which do this thing..
basically i want to rate product using no of positive reviews and no of negative reviews and sentiment score of a product.. if anyone know any algorithm please let me know..
You can add weightage to your parameters (might like to use multi-factor ranking where Positive and negative reviews are counted and given individual weights) and use simple formula as suggested in comments(assuming none of the parameters is 0), but this has a potential problem, products with 5 positive reviews and 1 negative will be ranked higher than 50 positive and 20 negatives.So, might be having info about how old is product(Age) and what is average positive review score per day will be a good idea. you can use Age for providing weight to reviews. See this link for not sorting by average rating
One suggestion is to create clusters based on sum of likes and dislikes ranges (you may like to use average positive reviews per day to change the clusters of products,This will rank new products with lot of positive reviews higher) and then use weighted algorithms, but it will require ranking two times ,firstly ranking within clusters,then ranking the clusters.