Rerank
I was frustrated when searching on Amazon that so many of the top products were full of suspicious reviews and I found myself disappointed with several 5-star purchases. I decided to write an algorithm that filters out suspicious reviews and provides a more accurate product ranking.
How it Works
ReRank generates an unbiased Amazon product ranking
by identifying the untrustworthy reviews and resorting the list of results
by identifying the untrustworthy reviews and resorting the list of results
Selects the Best Selling Products
We start with the top 30 ranked products and filter the list to only contain products that have at least 50 reviews. It is not fair that a product with two 4-star reviews is equal to a product with five-hundred 4-star reviews. This also helps weed out the products where a seller received such bad reviews that they simply relisted the product with a new name and started over.
Uses ReviewMeta Data
Review Meta is an awesome site that analyzes individual products for suspicious reviews, creates an extensive report on potential issues then generates a more accurate star ranking. I use their API to flag any products with an abnormal amount of suspicious reviews and remove them from my trusted list.
Rank Based on Negative Reviews
How many times have you seen a product with 80% 5-star reviews and 20% 1-star reviews but very few inbetween? An obvious indicator that something is not right. I created an algorithm to recalculate rankings based on the percentage of 1 star reviews.