An investigation of Social Support Exchange and Communication Patterns among Chinese on Online Discussion Forum

Xi Wang, Xin Tong, Yushan Zhu, Tianyi Tan, Bowen ZhengProceedings of the ... Annual Hawaii International Conference on System Sciences/Proceedings of the Annual Hawaii International Conference on System Sciences 2020 · DOI

Abstract

Online Health Communities (OHCs) are vital resources for patients worldwide. This study analyzes a representative Chinese breast-cancer-related OHC—the Baidu Discussion Forum—to investigate social support exchange and communication patterns through user-generated content. Using machine learning and social network analysis, we find that emotional support is the primary driver of interaction. However, the community structure is fragile, relying heavily on a small number of highly active "opinion leaders" and lacking the casual "companionship" talk common in Western forums. These findings have significant implications for the long-term sustainability of Chinese OHCs.

Introduction

Online Health Communities (OHCs) serve as global lifelines, yet most research has focused on Western populations. Cultural nuances, such as the collectivistic nature of Chinese society and specific stigmas surrounding cancer, can fundamentally alter how individuals seek and provide help. This study addresses whether communication patterns observed in Western OHCs generalize to a Chinese context, specifically examining the prioritization of information versus emotion and the role of "off-topic" social bonding.

Methodology

The research employed a two-pronged computational approach:

  1. Automated Text Classification: Machine learning models categorized posts into five types: Seeking/Providing Informational Support (SIS/PIS), Seeking/Providing Emotional Support (SES/PES), and Companionship (COM). Logistic Regression with Word2Vec embeddings proved most effective, achieving F1-scores between 0.82 and 0.89.
  2. Social Network Analysis (SNA): Five overlapping subnetworks were constructed based on support types. Topology and Jaccard coefficients were used to measure user-level overlap between different support categories.

The dataset comprised 186,291 posts from 36,184 users on the Baidu "Breast" forum over an 11-year span (2007–2018).

Key Findings

The Dominance of Emotional Support

Providing Emotional Support (PES) was the most frequent interaction type, accounting for 43.76% of all posts. In a culture where cancer is often a sensitive topic avoided in person, the OHC provides a safe outlet for emotional validation. As shown in the data, PES remains the dominant support type regardless of the longevity of a discussion thread, while companionship remains consistently low.

Distribution of social support types by thread longevity.

The "Opinion Leader" Phenomenon

The social network is heavily influenced by a tiny fraction of highly active users. For instance, one user contributed nearly 46,000 messages over five years. This concentration of activity creates a fragile network structure. The degree distribution highlights how a few "super-users" shape the connections, with sharp drops in probability at high degrees indicating the outsized influence of these individuals on the community's connectivity.

Log-log degree distribution of the social support networks.

The Companionship Gap

Unlike Western OHCs, where users often bond over non-health topics, Chinese users stay strictly "on-task." Companionship (COM) accounted for only 8.84% of posts. This transactional approach—getting help and leaving—may undermine the long-term community foundation, as there is less casual social bonding to sustain engagement over time.

Conclusion

Chinese OHCs are characterized by a strong emphasis on emotional support and a high reliance on a few key contributors. While effective for immediate needs, the lack of diverse social bonding and the vulnerability of the network to the departure of opinion leaders pose risks to community sustainability. Future efforts should focus on encouraging "middle-tier" users to ensure long-term resilience and community health.

Summary updated

Abstract

Online Health Communities (OHCs), frequently adopted as online discussion forums for online users to communicate on health issues, have been used worldwide. By analyzing a representative breast-cancer-related OHC from mainland China—Baidu Discussion Forum, this study attempts to investigate social support exchange and communication patterns through user-generated content by data mining approaches. According to the outcomes, emotional support seeking and providing presents itself to be a more critical theme among Chinese users than other types of social support. In addition, almost half of the users on Baidu Discussion Forum have simple patterns of involvement, and a fairly small proportion of highly active Chinese users are quite influential in shaping the connections of the social support network. Meanwhile, the off-topic discussions which are not directly on health concerns are not frequently touched by Chinese people. This may impact the longevity of both users and threads, and undermine the foundation of OHCs in the long term. The findings have practical implications for researchers and health practitioners targeting on the Chinese population.
BibTeX
@inproceedings{wang2020investigation,
  author = {Xi Wang and Xin Tong and Yushan Zhu and Tianyi Tan and Bowen Zheng},
  title = {{An investigation of Social Support Exchange and Communication Patterns among Chinese on Online Discussion Forum}},
  booktitle = {Proceedings of the ... Annual Hawaii International Conference on System Sciences/Proceedings of the Annual Hawaii International Conference on System Sciences},
  year = {2020},
  doi = {10.24251/hicss.2020.405},
}
Figure