Cyber Times International Journal of Technology & Management

In today’s era, due to high availability and affordability of Internet, large numbers of business opportunities have been generated. This leads to competition in the business world. The one who can retain their existing customers and attract new customer is the one who can survive in the market. To perform analysis on web services, web log files can be used. But due to large size of log files, it is practically impossible to use them directly. To overcome this problem web usage mining can be performed to extract previously unknown usage patterns. In this paper initially, basic concept of web usage mining will be discussed in details. The proposed approach and architecture of the tool will be discussed in the second section. The new approach integrates web usage mining with graph theory. Advanced technology: Neo4j which is used to generate and query graphs will be discussed in pattern mining phase. The future scope regarding management of huge log files using Hadoop are pointed out in the last part of the paper.