I'm not sure if stackoverflow is the right place to ask that but I need some suggestions on algorithms I could use for personalizing content (articles to be exact). I will collect information about a user from the articles he reads mostly. All the information collected will be converted to separate keywords with the count it is repeated. For example: keyword count
Computers 10
IT 12
Cars 2
I was thinking about Naive Bayes classification algorithm but the problem is that I don't really have data about unwanted articles.
Francesco is exactly right. This field is called recommender systems, and there's a very general categorization of collaborative versus content-based recommenders. Collaborative filtering uses the actions and opinions of other people to help you find items you'll like and avoid items you won't, whereas content-based filtering uses metadata about the content of items. The former needs a system of users in which you have some connection with other people (or items they've consumed); the latter requires a good database of metadata about items.