Product recommendation is about predicting ratings of users and creating personalized recommendations for users for a wide variety of products like books, clothes, movies, videos, or songs. Today this is mostly done online when users visiting Web sites or Web shops and the type of products can be basically anything. The only requirement that needs to exist is the data basis for these recommendations and rating predictions that are typically aggregated customer big data over a long period of time. Product recommendation engines apply statistical and knowledge discovery techniques on those previously recorded data in order to generate personalized product recommendations for a particular user. Much more details and the role of recommendations for e-Commerce can be found here.
The fundamental goal is to improve the conversion rate by pointing the customers to interesting products they want to buy. The recommendation helps customers thus to find products they most likely want to buy faster. In addition product recommendation promotes cross-selling by suggesting additional products. Also customer loyalty can be improved by creating a value-added relationship. Much more details on applications can be found here.
One of the first systems was called ‘Information Lense’ that was using a social filtering technique created in 1987 described in detail here. Since then product recommendation techniques evolved with recommender systems that are used by many market leaders in a wide variety of industries today. Examples are the Amazon shopping Web page, the Netflix movie rental system, or the online music recommendation system Pandora.
Product recommendation details
We recommend to watch the following video about this subject: