Label propagation algorithm based on adaptive H index

Abstract

Label propagation algorithm is a part of semi-supervised learning method, which is widely applied in the field of community partition. The algorithm is simple and fast, especially in the large complex community network. The algorithm shows nearly linear time complexity, but it has great instability and randomness. Many scholars make their improvements on the original label propagation, but most of them are not suitable for large community network discovery, which usually have higher time complexity. Therefore, we propose a label propagation algorithm based on adaptive H index, which improves the stability and accuracy of LPA by using the refined H index as a measure of node importance. Finally, the algorithm is tested by public standard dataset and synthetic benchmark network dataset, and the test result shows that the proposed algorithm has better stability and accuracy than some existing classic algorithms.