This paper introduces a comprehensive extension of the Inductive Pattern Search (IPS) algorithm to weighted bipartite graphs for profit-maximizing biclique mining. The proposed method, W-IPS, includes weight-aware partitioning strategies, profit-based pruning techniques, and modified termination conditions, achieving a worst-case time complexity of O(m·αn+n·βw). The approach is validated through experiments on e-commerce, financial fraud detection, and recommendation system datasets.
Key findings
W-IPS extends IPS algorithm to weighted bipartite graphs for profit-maximizing biclique mining.
Introduces weight-aware partitioning, profit-based pruning, and modified termination conditions.
Achieves a worst-case time complexity of O(m·αn+n·βw), preserving theoretical guarantees of IPS.
Demonstrates practical applicability and efficiency through experiments on real-world datasets.
Limitations & open questions
The paper does not discuss the scalability of the algorithm for very large graphs.
The effectiveness of the algorithm in dynamic environments is not addressed.