The difference between CF & personalization in the context of the web.

Collaborative filtering, also known as a recommendations system, is the method of making automatic predictions (filtering) about the interests of a user by collecting taste information from many users (collaborating). CF’s underlying assumption is “those who agreed in the past are likely to agree again in the future”. Those automatic predictions are specific to the user, but the information is collected from many users. Many different types of CF exist; some examples are active filtering, passive filtering, and item-based filtering. Some websites that use collaborative filtering include but are not limited to Amazon, NetFlix, TiVo and Musicmatch. It is demonstrated in Amazon when for example you are viewing or purchasing a book and then they advertise other books stating “customers who bought items in your recent history also bought...” This shows how Amazon used other customer’s collaborative information that is specific to the user by filtering his/her information (items purchased or viewed).
Web personalization is based on the individual’s interests. It is usually based on implicit data such as pages viewed and purchases made. If it were to use explicit data such as ratings and preferences or if the user was able to make changes within the website then it would be called customization. There are 'two types of personalization in the context of the web: rule-based and content-based. Rule-based personalization filtering is based on “if this-then that” rules processing as well as collaborative filtering. In Christina Ricci’s article “Personalization is not technology: using web personalization to promote you business goal”, she states that “web personalization is a strategy, a marketing tool, and an art”. She also said that:
“Personalization, properly implemented, brings focus to your message and delivers an experience that is visitor-orientation, quick to inform and relevant. Personalization, poorly implemented, complicates the user experience and orphans content”
From the way I see it, collaborative filtering and web personalization are extremely closely related. I find that collaborative filtering is a somewhat a form of personalization. However their biggest difference is that one derives information from like-minded individuals, the other from the user’s him or herself. Collaborative filtering has the benefit of being fast and more efficient in computation, allowing for quicker results and feedback. The disadvantage of collaborative filtering is that it may lead to less reliable recommendations. Like web 2.0 applications, the quality of recommendations improves with an increasing user population size. Personalization on the other hand may give more individualized recommendations but may also be limited by a lack of recommendations. Both should consider the ethical implications when designing their strategy so that they don’t infringe on the user or other user’s privacy. They are both necessary for building relations and acting as marketing tools therefore they should be carefully constructed with the best design and calculation architecture in order to optimize results.