TRANSACTIONS-FREQUENCY BASED GRADED LINK-CLASSIFICATION IN GRAPHS

Main Article Content

Ashish Gavande
Sushil Kulkarni

Abstract

Link classification categorises links between nodes of graphs for improved graph learning.
This work proposes a novel approach of using the frequency of transactions between nodes to learn
affinity for associations and thereby classifies links between nodes. Further, the classification is done
for multiple grades of classification and not just as strong/weak links. The model is successfully able
to classify links with around 95 percent micro-F1 accuracy on both homogeneous and heterogeneous
datasets using a multi-layer perceptron network.

Article Details

How to Cite
Gavande, A., & Kulkarni, S. (2022). TRANSACTIONS-FREQUENCY BASED GRADED LINK-CLASSIFICATION IN GRAPHS. INFOCOMP Journal of Computer Science, 21(1). Retrieved from https://infocomp.dcc.ufla.br/index.php/infocomp/article/view/1838
Section
Machine Learning and Computational Intelligence
Author Biography

Sushil Kulkarni, University of Mumbai

Associate Professor and Head, Dept. of Mathematics