Junghoo Cho

3Machine learning papers tracked by Lacuna
Is this you?

Research Overview

Junghoo Cho focuses on social network analysis, topic modeling, and the evaluation of user-generated content. His research trajectory spans the development of specialized topic models that account for entity popularity to the identification of topic-specific authorities on content sharing platforms. The most recent work in his profile investigates predictive modeling of microblogging behaviors using adaptive Bayesian approaches.

Research Trajectory

Cho’s research in the early 2010s addressed the unique challenges of applying traditional text mining techniques to the structured data of social networks. In standard natural language processing, highly frequent words are often discarded as meaningless stop words to reduce noise. However, in social networks, highly frequent nodes often represent popular individuals—such as politicians or celebrities—whose presence is central to the network's structure and meaning. As the senior author on Incorporating popularity in topic models for social network analysis (2013), Cho and his collaborators proposed a method to integrate popularity directly into topic models. This approach prevented frequent entities from introducing noise into identified topic groups without requiring their removal from the dataset.

Building on this interest in user roles and content quality, Cho's subsequent work examined how to identify domain experts within large-scale content sharing services like Flickr and YouTube. While previous systems often calculated a user's global authoritativeness, they struggled to differentiate expertise across specific subjects, which is critical given the highly variable quality of user-generated content. Acting again as senior author, Cho helped develop a principled methodology in Who are experts specializing in landscape photography?: analyzing topic-specific authority on content sharing services (2014) to discover topic-specific authorities among massive numbers of contributors.

In 2016, Cho co-authored Modeling a Retweet Network via an Adaptive Bayesian Approach, which shifted focus toward predictive modeling on microblogging platforms like Twitter. This research recognized that while many existing models analyzed original tweets, they often neglected the rich structural and behavioral signals embedded in retweet networks. By employing an adaptive Bayesian approach, the work demonstrated how incorporating secondary sharing information could significantly enhance predictive tasks such as tweet and followee recommendation.

Current Directions

Based on the most recent publications in the evidence from 2016, Cho’s research interests center on the predictive modeling of user behavior within microblogging environments. This involves leveraging advanced probabilistic frameworks, such as adaptive Bayesian approaches, to capture the complex dynamics of information propagation and user interaction. A primary focus in this domain is effectively integrating secondary behavioral signals—such as content sharing and network endorsements—to improve the accuracy of personalized recommendations and structural network analysis.

Profile updated

See something wrong? Report a mistake