Coincidence Complex Networks

TL;DR

This paper introduces a novel method for constructing complex networks using real-valued Jaccard and coincidence similarity indices, demonstrating improved performance and detailed modular structure detection in datasets like iris and C. elegans neuronal networks.

Contribution

The work applies new similarity indices to build and analyze complex networks, enhancing connectivity pattern detection over traditional cosine distance methods.

Findings

Enhanced network construction performance compared to cosine distance

Effective detection of modular structures in datasets

Successful application to biological and dataset-specific networks

Abstract

for representing, characterizing, and modeling an ample range of structures and phenomena from both theoretical and applied perspectives. The present work describes the application of the recently introduced real-valued Jaccard and coincidence similarity indices for building complex networks from datasets. More specifically, two nodes are linked whenever their similarity is greater than a given threshold. Weighted networks can also be obtained by taking the similarity indices as weights. It is shown that the proposed approach allows substantially enhanced performance when compared to cosine distance-based approaches, yielding a detailed description of the specific patterns of connectivity between the nodes. The impressive ability of the proposed methodology to emphasize the modular structure of networks is also illustrated with respect to the iris dataset and C.~elegans neuronal network connectivity data with remarkable results. The reported methodology and results pave the way to a significant number of possible theoretical and applied developments.