Abstract: The internet becomes a primary need or the part of daily life. A significant amount of growth of internet users is observed in recent years. In the similar ratio the internet frauds and phishing cases are also observed. In order to prevent these issues the user awareness is required. In addition of that various anti-phishing tools are also used to identify the phishing cases. There are a number of techniques available for detection of phishing URLs and prevent them to open in the user's browser. In this presented work the data mining technique based phishing URL detection technique is investigated. The data mining approaches are always first evaluating the historical data to recover the patterns to recognize and then the concluded knowledge is used for application. In the similar manner the phish tank data is used to recover the patterns and the mixed data is used for classification performance demonstration. In the proposed work first using the phishing URLs the significance of the URL is estimated. This phase is termed as a feature computation phase. After finding the features the entire URLs are encoded on the basis of these features and a transactional database is prepared. After this the association rule mining algorithms are applied to the dataset. In this experiment the Apriori algorithm and FP-Tree algorithm is used for computing the association rules. These association rules are further utilized for detection of phishing URLs. The implementation of this technique is performed on JAVA technology. After implementation the experimental results with increasing amount of data are performed. The result shows the increasing amount of data for classification impact on the performance of both the algorithms. But the FP-Tree algorithm provides efficient and accurate results as compared to the Apriori algorithm.
Keywords: phishing detection, URL analysis, FP-Tree, Apriori Algorithm, Data mining technique.