Product discovery via recommendation based on user comments

Recommendation systems on E-commerce websites help consumers to find products. A recommendation system learns consumer behavior in order to suggest products to those consumers. Recommendation systems allow consumers to have new experiences discovering new products rather than needing to search for them. When making purchase decisions consumers often use the comments left by previous buyers to help them. This paper presents how recommendation systems help E-commerce websites to recommend products, analyzes the recommendations used on some example sites and presents a new technique for recommendations based on the analysis of user comments and then analyzes the results of the new technique. The new techniques include parsing the text in comments to generate a word cloud based on the log likelihood of word frequencies, and then compares products using the RV Coefficient. Our approach automatically identifies similar products for recommendation, and based on the results of our experiment, the recommendations closely match those that would be manually chosen.