摘要
在推特(Twitter)上抓取10万余条关于"中国文化"的帖子及其发布者信息,以及收藏或点赞该帖子的用户样本信息,并将用户所属区域按照国家或地理区位划分出39个版块。在对数据进行清洗后留下39行×39列的传受连通关系,根据不同信息版块用户的发布、点赞和收藏情况,运用社会网络分析法确定其在全球传播中的位置。同时,采用LDA(Latent Dirichlet Allocation)主题模型计算向量夹角的余弦值(余弦距离),刻画不同版块用户的信息内容相似度。最后得出某版块用户所处的社会网络位置与该版块和其他版块的平均信息距离呈现显著正相关关系的结论。
More than 100 thousand posts about"Chinese culture"on twitter were captured and the information of their publishers and people who like or collect the posts were obtained. The users' regions were divided into 39 information blocks according to geographical locations,leaving a 39 × 39 matrix with transmitter-receiver relationship to locate users' position in global communication based on their situation in different information blocks by means of SNA( social network analysis). Meanwhile,LDA( latent dirichlet allocation) model was used to observe the probability distribution of 100 thousand posts on 50 topics,from which we can understand the similarity of "Chinese culture"description from different blocks by calculating the cosine of the vectors to get an information distance. The results show that the average information distance of users from different blocks is positively correlated with their social network positions,in which"intermediary role"based on social connection frequency and degree of social interactive connection plays an important role.
引文
[1]斯科特.社会网络分析法[M].刘军,译. 2版.重庆:重庆大学出版社,2007:6-21.
[2]蔡骐,陈月.基于社会网的知乎网意见领袖研究[J].湖南师范大学社会科学学报,2018(5):128-138.
[3]陈静.知乎用户网络与意见领袖在舆情传播中的作用[J].新闻研究导刊,2017,8(8):66-67.
[4]廖圣清,黄文森,易红发,等.群体性事件中参与主体的策略演化博弈分析———一种社会网络视角[J].新闻大学,2016(6):87-99+150.
[5]石义彬,吴世文.我国大众传媒再现和建构中国文化身份研究———基于数字传播和全球传播环境的思考[J].当代传播,2010(5):4-7.
[6]吴瑛,李莉,宋韵雅.多种声音一个世界:中国与国际媒体互引的社会网络分析[J].新闻与传播研究,2015,22(9):5-21+126.
[7]韦路,丁方舟.社会化媒体时代的全球传播图景:基于Twitter媒介机构账号的社会网络分析[J].浙江大学学报(人文社会科学版),2015,45(6):91-105.
[8]姜鑫.社会网络分析方法在图书情报领域的应用研究[M].北京:知识产权出版社,2015:58-77+105-109.
[9]刘军.整体网分析———UCINET软件实用指南[M]. 2版.上海:上海人民出版社,2014:254-263.
[10]伯特.结构洞:竞争的社会结构[M].任敏,李璐,林虹,译.上海:上海人民出版社,2008:1-25.