While buying something from Amazon.com, many of us probably have seen the section called “Customers Who Bought This Item Also Bought”. This is the famous recommender system. Amazon was granted a patent in 1998 for this “collaborative recommendations using item-to-item similarity mappings”. This technique has now become ubiquitous. Most of the e-commerce sites currently offer recommendations based on customers’ purchase history. Even the matrimonial sites took up this concept. Alongside the profile you are currently viewing, the sites display a handful of additional profiles under the heading, “Those who viewed this profile also viewed the following profiles”.
This is how it works. Suppose you are buying the book ‘Never Let Me Go’ by Kazuo Ishiguro. The system checks which other customers had bought this same book. Then, it finds out what other books those customers had bought in the past. If the current buyer is yet to buy those books, they appear in the recommended list. As evident, the process works based on a large pool of consumer behavior data, rather than any sophisticated algorithm.
The precision and usefulness of the system increase as the customer base and individual purchase frequency increase. In addition, the more information you share (by way of mentioning your preferences, rating a purchase etc) the more relevant recommendations you get.