All Friends are not Equal: Weight-Aware Egocentric Network-Based User Profiling

摘要

Several studies were interested in social shared content as rewarding source to enrich user's profile. In this context, we highlight the importance of considering the user's relationship strength to derive relevant social profile. We focus on egocentric network community-based user profiling process (CoBSP) which considers only binary connections between people. In real life, social connections are not merely binary but have associated weights that record people's relation strength. We aim to enhance the effectiveness of CoBSP process by integrating ties strength in the user's profiling process. Our motivation stems from the intuition that people having the strongest relationships with the profiled user may reveal more valuable information about him. The first experiments conducted on scientific publications networks (DBLP/ResearchGate) are promising and show the relevance of our prior premise. However, we should conduct further experiments to completely validate the effectiveness of our proposition.