多平台视角下用户知识交流主题挖掘与画像分析——以ChatGPT话题为例  

User Knowledge Exchange Topic Mining and Portrait Analysis from Multiple Platforms——A Case Study of ChatGPT

作　　者：严炜炜[1] 曹灿瑜 Yan Weiwei;Cao Canyu(School of Information Management,Wuhan University,Wuhan 430072,China)

机构地区：[1]武汉大学信息管理学院,湖北武汉430072

出　　处：《现代情报》2024年第7期47-59,共13页Journal of Modern Information

基　　金：国家自然科学基金面上项目“情境意识驱动的跨平台知识交流行为及其价值共创研究”(项目编号:72374159);中央高校基本科研业务费专项基金资助项目“多元社区情境下用户知识交流价值识别与共创研究”(项目编号:2042023kf0173)。

摘　　要：[目的/意义]基于多平台视角挖掘用户知识交流主题特征,构建用户画像标签体系,有助于全面理解用户利用差异化平台开展知识交流行为的规律偏好,为平台提供精确优化策略,为平台间生态建设提供参考依据。[方法/过程]采集大众社交平台、兴趣交流平台、垂直知识平台3类典型平台中ChatGPT话题相关的原创博文及用户数据,采用BERTopic模型凝练知识交流主题,结合多平台数据特点,从自然属性、社会属性、知识交流行为属性和知识交流主题属性4个维度抽出画像标签,通过K-means聚类实现用户画像,呈现群组特征并进行平台差异对比。[结果/结论]研究识别出了应用场景、行业进展、未来探讨、相关产业、咨询求助、热门话题、使用感受、风险监督8大知识交流方向及46个主题;根据属性特征将用户划分为专业贡献型、综合共享型、社交求知型和话题潜力型4类,平台间知识交流主题和用户画像存在显著差异,各平台应采取差异化的激励方式,增强平台用户黏性。[Purpose/Significance]The study aims to understand users preferences of knowledge exchange on different platforms can be achieved by analyzing the topics discussed across multiple platforms,and construct a user portrait labeling system,which can help platforms provide more personalized optimization strategies and improve interplatform ecological construction.[Method/Process]The study collected original blog posts and user data related to ChatGPT topics from three typical platforms:mass social platform,interest exchange platform,and vertical knowledge community.Next the BERTopic model was adopted to condense knowledge exchange topics.Then,the study extracted portrait labels from four dimensions:natural attributes,social attributes,knowledge exchange behavioral attributes,and knowledge exchange thematic attributes.To realize user profiles,present group characteristics,and compare platform differences,the study used K-means clustering to implement a user portrait labeling system.[Result/Conclusion]The study has identified 46 different topics and directions for knowledge exchange on 8 major cutting-edge science and technology subjects.These subjects include application scenarios,industry progress,future exploration,related industries,consultation and help,hot topics,experience of use,and risk supervision.Furthermore,the study has classified the users into 4 categories based on their attribute characteristics.These categories are professional contribution,comprehensive sharing,social knowledge-seeking,and topic potential.The study also finds significant differences between platforms in terms of knowledge exchange topics and user profiles.Therefore,platforms should adopt differentiated incentives to enhance platform user stickiness.

关 键 词：知识交流 多元平台 主题模型 BERTopic 用户画像 

分 类 号：G203[文化科学—传播学]