follows a brief analysis of several typical recommendation engine applications, where two areas are selected: Amazon as a representative of e-commerce and watercress as a proxy for social networking.
recommendation in e-commerce applications – Amazon
Amazon as the originator of the recommendation engine, it has introduced the idea of penetration in every corner of the application. The core of the Amazon recommendation is to compare the user preferences to other users through data mining algorithms to predict the items that the user might be interested in. All corresponding to the above described recommendation mechanism, Amazon is used in the mixed partition mechanism, and different results in different area of recommendation is displayed to the user, figure two below shows the user can get in on the Amazon recommendation.
Amazon recommendation mechanism – home page
Amazon recommendation mechanism – browse items
Amazon uses the behavior that can record all users on the site, according to the characteristics of different data to deal with them, and divided into different areas for users to push recommendation:
today recommended (Today s Recommendation For You): usually based on the user’s recent history to buy or view records, and combined with popular items, given a compromise recommendation.
new product recommendation (New For You): the adoption of a content based recommendation mechanism (Content-based Recommendation), the new items to recommend to the user. Because the new item does not have a large number of user preferences, the content based recommendation can solve the problem of "cold start".
(Frequently Bought Together) bundling: analysis using data mining technology to find the user’s buying behavior, often with the same person or buy goods, bundling, this is a collaborative filtering recommendation mechanism based on a typical project.
people to buy / browse commodity (Customers Who Bought/See This Item Also Bought/See): This is a collaborative filtering recommendation application project based on typical, through social mechanism users faster and more convenient to find the interested items.
is worth mentioning, Amazon in doing recommendations, the design and user experience has done particularly unique: