A discriminative method to conduct exploratory analyses in networks

TL;DR

This paper discusses a discriminative approach for exploratory network analysis but has been withdrawn due to a misinterpretation of the original framework by Newman and Leicht.

Contribution

It attempted to clarify and build upon Newman and Leicht's mixture models for network analysis, but was retracted for conceptual inaccuracies.

Findings

The paper was withdrawn due to misinterpretation.

It acknowledged the correct interpretation of the original model.

The authors expressed gratitude for the clarification from Newman and Leicht.

Abstract

This paper, which commented on Newman and Leicht's "Mixture models and exploratory analysis in networks" (2007, PNAS 104, 9564-9569), has been withdrawn. The reason for this removal is that we misinterpreted the conceptual framework that the authors of the original paper use. Specifically, it is assumed in our paper that the variable theta_ri denotes the *a priori* probability that there exists an edge from group r to vertex i. The correct interpretation is that theta_ri denotes the probability that a given edge from group r connects to vertex i. We are very grateful to Mark Newman and Elizabeth Leicht not only for pointing out our misinterpretation, but also for doing it so politely and gracefully.