Top-K Similar Access Behavior Based On Structural Equivalence

Abstract

The number of online business transactions and Social networks on World Wide Web are rapidly increasing and producing enormous amounts of data. Mining and analyzing relevant data from the web log data for user's navigational behavior has a great significance in the area of data mining. In this paper, we aim to find top-k similar access behavior for an IP_address-Product network in the context of object-to-object relationship. We define similar behavior of users through distinct IP addresses about different categories in the context of node behavior of a bipartite network. As a part of user navigational behavior, we present (i) Top-k similar (Toks) algorithm which is the baseline to mine at most top-k IP addresses for each product category, and an efficient algorithm, (ii) A Recursive Approach for Top-k Similar IP addresses (RAFTS), which mines top-k IP addresses for each and every product category based on method of early associates. We run both the algorithms on four distinct networks and compare performance in terms of running time and memory. The algorithm RAFTS performs better than the baseline algorithm regarding running time and memory consumption. The above algorithms are useful to forecast the similar visiting behavior of the users through distinct IP addresses about different categories. This leads to improve the quality of products by identifying less influence able products.