Recommender System

Table of Contents:

Definition

Recommender (someone or something that recommend {Advising or suggesting that someone or something is the better option for a role or purpose.}) system (A three part system that provide an input, uses a transformation and expect an output) A recommender system is used to generate appropriate recommendations of items and products that might be useful to users. This is based on previous activities.

Description

Recommendation system gives information in one of two ways By Collaborative filtering and content base filtering.

Collaborative Filter

This is done by using a previous items purchased and selected by the user to give them suggestions. Decisions that are made by other users that are similar are also used to give suggestions to the user.

Content Based Filter

This uses distinctive characters of the user to predict more items with similar properties.
The Differences in the two systems can be found from looking at Last.fm and Pandora. Radio. Pandora uses a content bases filter while Last.fm uses a collaborative filter.

Benefits

• It drives Traffic to the site by using personalised email messages
• It turns customers into shoppers by giving them the user recommendations that they might be interested in.
• The shoppers are engaged with the site because the information they receive is personalised.
• It provide relevant document that target the need of an individual rather than the general public.

Negatives

• It requires a large number of data
• It is not easy to identify user changing preference once a preferences as been established

Conclusion

Being able to get recommendation that is targeted at individual on the web is crucial because this uses a strategic method that is being adapted in the supermarkets.

Reference

http://www.thefreedictionary.com/Recommender
http://www.vikas.sindhwani.org/recommender.pdf
http://en.wikipedia.org/wiki/Recommender_system
http://readwrite.com/2009/01/28/5_problems_of_recommender_systems

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License