Enhancing user reliability using a contextual-based approach on heterogeneous graph

The emergence of social media platforms like Reddit has significantly influenced financial markets, where online communities now shape individual behaviors and investment decisions. Identifying influential users within these networks poses challenges due to the vast amount of data generated and the complexity of defining influence. This paper presents a novel approach that integrates social network metrics with sentiment analysis to quantify user influence more effectively. We introduce the Content-based Centrality score (CbC), a metric combining user engagement, sentiment derived from posts, and traditional centrality measures. Using real data from Reddit's financial communities, we compare the CbC to established social media centrality metrics and validate its effectiveness through a human evaluation process, assessing the relevance and trustworthiness of financial posts. Our results demonstrate that the CbC provides a more nuanced and accurate understanding of user influence, offering valuable insights into the role of social media in shaping financial market behaviors.