Uncovering How Information Quality Shapes Diverse User Engagement on Content Community Platforms: Harnessing Deep Learning for Feature Extraction

The success of content community platforms (CCPs) heavily depends on the active engagement of users attracted by externally generated content. Previous research has highlighted the differentiation among various forms of user engagement, such as likes, comments, and retweets, in shaping the dynamics of value co-creation on CCPs. Our objective is to uncover distinct patterns of impact that information quality features have on these different forms of user engagement. Specifically, we employed deep learning techniques to extract information quality features and identified them as persuasive factors operating through central and peripheral routes based on the elaboration likelihood model (ELM), stimulating user engagement. Our dataset was derived from MaBeeWoo, China’s largest specialized CCP for travelogues with minimal barriers for creating text and image-based travelogues. By utilizing a negative binomial model, our analysis reveals significant differences in antecedents between retweets and likes/comments while also highlighting variations in the impact levels of specific content quality features between likes and comments. These findings suggest contrasting patterns regarding how content quality features influence information production and dissemination on CCPs, underscoring the necessity for platform sponsors to develop adaptive mechanisms aligned with their strategic objectives for incentivizing specific quality features.