Edge Weight Prediction in Weighted Signed Networks

by Srijan Kumar, Francesca Spezzano, V.S. Subrahmanian and Christos Faloutsos


Photo

Predicting the reliability of users in any network is important in challenging. In this work, we assign scores to each user and each recipient to measure their reliability and quality. We create two metrics for each node: fairness to measure the reliability of users while giving ratings and goodness to measure their quality while being rated. Both metrics are interdependent. The intuition behind the metrics is the following:
If a user consistently rates a 5 star-deserving product as 5 star and 1 star-deserving product as 1-star, then the user is fair. And a high quality (good) product gets high ratings from highly fair users
In this paper, we formulate users' fairness and goodness in terms of each other. We use these scores to predict the weights of hidden ratings.

Paper

Download the PDF

Slides for the conference presentation at ICDM 2016.

Abstract
Weighted signed networks (WSNs) are ones in which edges are labeled with positive and negative weights. WSNs can capture like/dislike, trust/distrust, and other social relationships between people. In this paper, we consider the problem of predicting the weights of edges in such networks. We propose two novel measures of node behavior: the goodness of a node intuitively captures how much this node is liked/trusted by other nodes, while the fairness of a node captures how fair the node is in rating other nodes' likeability or trust level. We provide axioms that these two notions need to satisfy and show that past work does not meet these requirements for WSNs. We provide a mutually recursive definition of these two concepts and prove that they converge to a unique solution in linear time. We use the two measures to predict the precise edge weight in WSNs. Furthermore, we show that when compared against several individual concepts used in both the signed and unsigned social network literature, our fairness/goodness (FG) measures almost always show the best predictive power. We then use these as features in different multiple regression models and show that we can predict edge weights on 2 Bitcoin WSNs, an Epinions WSN, 2 WSNs derived from Wikipedia, and a WSN derived from Twitter with far more accurate results than past work. Moreover, FG measures form the most significant feature for prediction in most (but not all) cases.

Bibtex

@inproceedings{kumar2016wsn, author = {Kumar, Srijan and Spezzano, Francesca and Subrahmanian, V.S. and Faloutsos, Christos}, title = {Edge Weight Prediction in Weighted Signed Networks}, booktitle = {Data Mining (ICDM), 2016 IEEE International Conference on}, year = {2016} }

DATA and code

The weighted signed networks data used in the paper can be downloaded below!

Download Data

The code of the Fairness and Goodness algorithm be downloaded below!

Download Code