the ongoing increase in the usage of web has led to accumulation of large amounts of data every second. This has in turn made the research industry to grow and focus towards employing web usage mining for increasing the revenues for businesses, carrying out analysis on browsing behavior of web users, improving website layout and much more. Web usage mining is becoming increasingly popular due to the huge amounts of benefits it offers. The most significant amount of information about the web usage by the users is contained in log files usually called as server log files. The clicks made by the user, the order in which they are made and the time spent on each web page is all contained in these logs and is referred to as clickstream data or clickpath data. An attempt has been made through this paper to provide a holistic view as to what clickstream data analysis is, how mining techniques are applied on such data to generate useful information and what kind of applications exploit it to get useful information.

Web usage data is extensively used in every domain to analyze the browsing behavior of the users who visited the website or search engines. Web usage data is contained in server logs called as Web Logs. This data enables the website owners to infer the needs and interests of the users for using this information to increase the revenue from their web business. The website owners employ recommender systems for this purpose. The recommender systems exploit web usage data to predict what web pages the user will visit next and therefore offer the recommendations for those very pages to the user and offers them support while browsing. This in turn helps users to have a better browsing experience, personalized support and hence, probability of user buying out the products from that website increases. Web usage mining alone is used in traditional recommender systems. Modern recommender systems employ semantic knowledge base i.e. domain knowledge in addition to web usage mining for efficient prediction of pages as this helps in avoiding the new page problem. This paper presents a comparative and comprehensive study of modern and traditional recommender systems.