[目的/意义] 在线健康社区中对高影响力用户的有效识别，有助于健康信息需求者发现有价值的健康信息，对于降低健康信息查找成本和提高健康行为决策的有效性具有重要意义。[方法/过程] 从用户交互性和评论情感倾向出发，利用PageRank和SVM等算法构建出在线健康社区用户影响力的测量方法，并以医享网为实验对象，从发布内容使用价值的视角，进一步计算了该社区中用户的综合影响力，并对案例用户进行分析。[结果/结论] 分析结果表明该算法具有一定的合理性，能够对PageRank算法的影响力计算结果进行优化；同时，利用TF-IDF和互信息算法揭示了高综合影响力用户发布的信息内容与社区其他用户群体内容主题基本一致，该类用户对社区的主题方向起到一定的引导作用。因此，通过本研究所构建的方法可以有效识别高影响力的用户，有助于健康信息需求者及时准确的发现所需信息，提高健康信息的使用效果，从而丰富在线健康社区用户信息行为的理论和实践研究。
[Purpose/significance] The effective identification of high-impact users in online health communities is helpful for demanders to find valuable health information, which is of great significance for reducing the cost of health information search and improving the effectiveness of health behavior decision-making. [Method/process] This study was from the perspective of interactivity of users and emotional tendency of comments, using PageRank and SVM algorithm to build a method to measure the users’ influence in online health community, and took the medical network as experimental object, from the angle of content use value, further calculated the comprehensive influence of users in the community, and in case the user is analyzed. [Result/conclusion] The results show that the algorithm is reasonable and can optimize the calculation results of PageRank algorithm. At the same time, the TF-IDF and Mutual Information algorithm are used to reveal that the information content published by high comprehensive influence users is basically consistent with content topics of other user groups in the community, and such users play a certain role in guiding the theme direction of the community. Therefore, the method constructed in this study can effectively and reasonably identify high-impact users, which is helpful for health demanders to find the required information timely and accurately, improving the effect of using health information, so as to enrich the theoretical and practical research on the information behavior of users in online health communities.
[1] 杨梓. 最新！卫健委发布全国医疗相关数据[EB/OL].[2021-04-27].https://www.sohu.com/a/247593213_439958.
[2] 2018年中国健康医疗大数据行业发展现状及发展趋势分析[EB/OL].[2021-04-27].http://www.chyxx.com/industry/201806/649591.html.
[3] 国务院办公厅. 国务院办公厅关于促进"互联网＋医疗健康"发展的意见[EB/OL].[2021-04-27].http://www.pkulaw.cn/fulltext_form.aspx?Db=chl&Gid=37395b41f6f018e4bdfb&keyword=%E5%8C%BB%E7%96%97&EncodingName=&Search_Mode=accurate&Search_IsTitle=0.
[4] 王佳敏,吴鹏,陈芬,等.突发事件中意见领袖的识别和影响力实证研究[J].情报学报, 2016, 35(2):169-176.
[5] 赵发珍.基于链接分析法的网络社区影响力研究——以国内30个网络社区网站为例[J].现代情报,2013,33(6):91-95.
[6] 董伟,李建红,陶金虎.在线健康社区活跃用户识别及其交互类型分析[J].文献与数据学报,2020,2(1):89-101.
[7] 陈远, 刘欣宇. 基于社会网络分析的意见领袖识别研究[J]. 情报科学, 2015, 33(4): 13-19, 92.
[8] 谢英香,冯锐.虚拟教师社区中博客网络位置的影响力研究[J].现代教育技术,2010,20(1):97-100,110.
[9] JONNALAGADDA S, PEELER R, TOPHAM P. Discovering opinion leaders for medical topics using news articles[J]. Journal of biomedical semantics,2012,3(1):2.
[10] 陈芬,高小欢,彭玥,等.融合文本倾向性分析的微博意见领袖识别[J].数据分析与知识发现,2019,3(11):120-128.
[11] 刘玲,杨长春.一种新的微博社区用户影响力评估算法[J].计算机应用与软件,2017,34(7):212-216,261.
[12] 张俊豪,顾益军,张士豪.基于PageRank和用户行为的微博用户影响力评估[J].信息网络安全,2015(6):73-78.
[13] SONG X, CHI Y, HINO K, et al. Identifying opinion leaders in the blogosphere[C]// ACM. Proceedings of the sixteenth ACM conference on conference on information and knowledge management. Lisbon: ACM, 2007:971-974.
[14] 韩忠明,苑丽玲,杨伟杰,等.加权社会网络中重要节点发现算法[J].计算机应用,2013,33(6):1553-1557,1562.
[15] 肖宇, 许炜, 商召玺. 微博用户区域影响力识别算法及分析[J]. 计算机科学, 2012, 39(9): 38-42.
[16] 马凤.基于PageRank算法的期刊影响力研究[J].情报杂志,2014,33(12):103-108.
[17] ZHANG Y. Determinants of poster reputation on internet stock message boards[J].American journal of economics and business administration,2009,1(2):114.
[18] Precision,Recall,F1score,Accuracy的理解[EB/OL].[2021-04-27]. https://blog.csdn.net/u014380165/article/details/77493978.
[19] ALAM S, YAO N. Big data analytics, text mining and modern english language[J]. Journal of grid computing, 2019,17(2): 357-366.
[20] 费洪晓,康松林,朱小娟,等.基于词频统计的中文分词的研究[J].计算机工程与应用,2005(7):67-68,100.
[21] 董伟,陶金虎.基于主题偏好的在线健康社区用户兴趣群体识别研究——以医享网为例[J].情报科学,2021,39(3):88-93,119.
