Recommender System is a wide area that has many sub fields that require a deep understanding and great research efforts. In particular the main aspects are: information inputs that are used by the algorithm that impacts the recommendations, algorithms that are hidden background and run the recommendation engine to predict the user¿s preferences, evaluation metrics that defines the satisfaction of the user and the quality of the recommendations.The sole dependency on user profile based on navigation history alone cannot promise the quality of recommendations in terms of accuracy and diversity because of lack of semantics in the processing. The time parameter in recommender systems should be considered on top of conceptual semantics as it has a great influence on item¿s popularity and user¿s preferences. The traditional evaluating metrics could not able to deal cold-start problem, that occurs with new users and new or less popular items in the web domain, because of the traditional filtering methods that mix up all users and items with same intent.