Mining of Web Server Logs using Extended Apriori Algorithm

Association rule mining is one of the most significant techniques in the field of data mining. It is very useful in discovering relationships hidden in large transaction datasets such as frequent patterns, associations etc. One of the popular and important algorithms in this category is Apriori algorithm which finds frequent itemsets using an iterative approach. But it suffers from a major limitation that in case of large databases, it requires a large number of passes while searching the frequent itemsets, thus increasing its scanning time. In order to lessen this time, an improved version of Apriori algorithm, called Extended Apriori is proposed in this paper which decreases the number of transactions in the database, hence reducing size of the database so as to minimize the scanning time. This extended algorithm is then used to mine web server logs of an educational web site in order to discover frequently visited pages by the user and also its performance is compared graphically with existing Apriori algorithm. In the end, the paper also outlines some future research directions in the area of web server log mining.