图书情报知识 ›› 2025, Vol. 42 ›› Issue (3): 145-158.doi: 10.13366/j.dik.2025.03.145

• 知识、学习与管理 • 上一篇    

在线健康社区知识共创的群体共识演化模式研究

易明, 熊雨彤, 刘明, 周阳   

  1. 华中师范大学信息管理学院,武汉,430079
  • 出版日期:2025-05-10 发布日期:2025-06-23
  • 通讯作者: 易明(ORCID: 0000-0002-4864-6025),博士,教授,研究方向:信息行为与个性化服务,Email: yiming0415@ccnu.edu.cn。
  • 作者简介:熊雨彤(ORCID: 0009-0001-9181-416X),硕士研究生,研究方向:用户行为与信息服务,Email: xyt0_0@163.com;刘明(ORCID: 0009-0007-9530-1322),博士研究生, 研究方向:个性化推荐,Email: isliuming@163.com;周阳(ORCID: 0000-0001-7221-5535),博士研究生,研究方向:用户信息行为,Email: 1850977851@qq.com。
  • 基金资助:
    本文系国家社会科学基金重点项目“在线健康社区知识共创机理及引导机制研究”(21ATQ006)。

The Evolutionary Patterns of Group Consensus for Knowledge Co-creation in Online Health Communities

YI Ming, XIONG Yutong, LIU Ming, ZHOU Yang   

  1. School of Information Management, Central China Normal University, Wuhan, 430079
  • Online:2025-05-10 Published:2025-06-23
  • Contact: Correspondence should be addressed to YI Ming, Email: yiming0415@ ccnu.edu.cn, ORCID: 0000-0002-4864-6025
  • Supported by:
    This is an outcome of the Key Project "Research on the Mechanism of Knowledge Co-creation and Guidance Mechanism of Online Health Community"(21ATQ006)supported by National Social Science Foundation of China.

摘要: [目的/意义]在线健康社区知识共创的本质是通过群体认知和群体共识的逐步提升而不断完善疾病治疗方案的过程。本文旨在从宏观及微观层面分别揭示群体共识演化模式特征及其关键影响因素,进而为引导和干预知识共创进程提供依据。[研究设计/方法]以“与癌共舞”社区2013年1月—2023年8月的998个研讨帖中的16,822条有效发言数据为基础数据集。首先针对用户立场的隐蔽性,提出融合语义结构分析的群体共识测度方法;其次在此基础上结合K-mediods和DTW算法提炼群体共识的演化模式;最后构建基于LightGBM算法的预测模型,从而对在线健康社区群体共识演化模式进行预测分析。[结论/发现]在线健康社区知识共创的群体共识演化模式可以抽象为高起低趋、大幅波动、突增渐缓和阶段递增四类。消极情感得分、议题质量、可信度等七种因素对在线健康社区群体共识演化模式的形成影响较大,其中议题质量和社会属性相似度对促进群体共识增长的作用尤为显著。针对高起低趋型研讨帖,建议通过采用简明的议题表述并鼓励相似背景用户参与讨论来提高群体共识。[创新/价值]提出了适用于在线健康社区知识共创的群体共识测度方法,并为群体共识的动态研究提供了新的拓展思路。

关键词: 在线健康社区, 知识共创, 群体共识, 演化模式, 模式预测

Abstract: [Purpose/Significance] The essence of knowledge co-creation in online health communities(OHCs)lies in the improvement of disease treatment approaches through the gradual enhancement of group cognition and the formation of group consensus. This study aims to reveal the characteristics of group consensus evolution patterns and their key influencing factors at both macro and micro levels, thereby providing a basis for guiding and intervening in the knowledge co-creation process. [Design/Methodology] This study uses a data set comprising 16,822 valid comments extracted from 998 discussion posts in an OHC called "Yuaigongwu", spanning from January 2013 to August 2023. Firstly, to address the concealment of users' standpoint, a group consensus measurement method integrating semantic structure analysis is proposed. Secondly, on this basis, the evolution patterns of group consensus are identified and refined by employing the K-mediods and Dynamic Time Warping(DTW)algorithm. Finally, a predictive model based on the LightGBM algorithm is constructed to analyze and forecast the evolution patterns of group consensus in online health communities. [Findings/Conclusion] The research results show that the evolution patterns of group consensus in knowledge co-creation within OHCs can be abstracted into four categories: Initial Surge-Decay Pattern, Significant Fluctuation Pattern, Swift Surge-Stabilization Pattern, Stage-wise Incremental Pattern. Seven factors, such as negative emotion scores, topic quality, and credibility, have a relatively significant influence on the formation of group consensus evolution patterns in online health communities. Among them, topic quality and social attribute similarity play a particularly prominent role in promoting group consensus growth. For discussion posts with a Initial Surge-Decay Pattern, it is recommended to enhance group consensus by adopting concise topic expressions and encouraging participation from users with similar backgrounds. [Originality/Value] This study introduces a method for measuring group consensus that is applicable to knowledge co-creation in OHCs, and provides new expansion ideas for developing dynamic research on group consensus.

Keywords: Online health communities, Knowledge co-creation, Group consensus, Evolution pattern, Model prediction