An Optimized Weighted Association Rule Mining On Dynamic Content

TL;DR

This paper introduces an optimized weighted association rule mining approach for dynamic content, utilizing an enhanced HITS algorithm with online eigenvector computation to efficiently handle frequent updates in transaction data.

Contribution

It develops a novel weight assignment method based on a directed graph and enhances the HITS algorithm for dynamic, real-time association rule mining.

Findings

Enhanced HITS algorithm is more efficient than the original in dynamic data contexts.

The method effectively updates item importance in response to frequent data changes.

The approach improves the accuracy of weighted association rules in dynamic environments.

Abstract

Association rule mining aims to explore large transaction databases for association rules. Classical Association Rule Mining (ARM) model assumes that all items have the same significance without taking their weight into account. It also ignores the difference between the transactions and importance of each and every itemsets. But, the Weighted Association Rule Mining (WARM) does not work on databases with only binary attributes. It makes use of the importance of each itemset and transaction. WARM requires each item to be given weight to reflect their importance to the user. The weights may correspond to special promotions on some products, or the profitability of different items. This research work first focused on a weight assignment based on a directed graph where nodes denote items and links represent association rules. A generalized version of HITS is applied to the graph to rank the items, where all nodes and links are allowed to have weights. This research then uses enhanced HITS algorithm by developing an online eigenvector calculation method that can compute the results of mutual reinforcement voting in case of frequent updates. For Example in Share Market Shares price may go down or up. So we need to carefully watch the market and our association rule mining has to produce the items that have undergone frequent changes. These are done by estimating the upper bound of perturbation and postponing of the updates whenever possible. Next we prove that enhanced algorithm is more efficient than the original HITS under the context of dynamic data.