A Method of Opinion Leader Discovery Based on Comprehensive Influence and Sentiment Characteristics

doi:10.11871/jfdc.issn.2096-742X.2021.04.011

Abstract

Abstract:

[Objective] In response to the phenomenon that traditional opinion leader discovery methods are limited to partial data features and lead to ignoring of some opinion leaders, this paper proposes a new opinion leader discovery method called CI-SC to achieve the purpose of discovering the ignored opinion leaders. [Methods] CI-SC integrates the influence characteristics of users in both profile attributes and post interaction behaviors to build a comprehensive influence evaluation index. By introducing users' sentiment characteristics, CI-SC achieves the discovery of opinion leaders through cluster analysis based on the above characteristics. The results are evaluated by analyzing identity information and social connections. [Results] According to the experimental results, the opinion leaders found by our method have higher statistical significance and lower overlap ratio compared with those found by traditional methods, which means that by considering more aspects of information, our method can effectively identify opinion leaders with greater influence and obvious sentiment characteristics that are ignored by traditional methods. [Limitations] The proposed method has only been tested on a small dataset so far. Thus, more experiments on larger datasets are needed to further validate its effectiveness. [Conclusions] This paper proposed a new opinion leader discovery method called CI-SC, which is based on comprehensive influence and sentiment characteristics. Experimental results prove that by taking into account the influence of a user in different aspects and further combining the influence with the sentiment characteristics, CI-SC improves the traditional opinion leader discovery methods in discovering opinion leaders without neglecting some information. Therefore, our method can be an effective supplement to traditional methods and has practical value in opinion analysis and guidance.

Key words: electronic invoice, block chain, duplicate reimbursement, information immutability, CI-SC, influence, sentiment polarity, opinion leader discovery

WANG Jiaqi,DU Yihua,ZHAO Yixia. A Method of Opinion Leader Discovery Based on Comprehensive Influence and Sentiment Characteristics[J]. Frontiers of Data and Computing, 2021, 3(4): 126-139.

Figures/Tables 11

Fig.1

Fig.2

Table 1

Table 2

Fig.3

Table 3

Table 4

Fig.4

Fig.5

Fig.6

Fig.7

References 32

[1]	Sendall P, Ceccucci W, Alan R Peslak. Got Web 2.0? A review of Web 2.0 tools for the information systems curriculum[J]. Information Systems Education Journal, 2010,8(28):1-14.
[2]	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.
[3]	Lazarsfeld P F, Berelson B, Gaudet H. The People's Choice[M]. Oxford, England: Duell, Sloan & Pearce, 1944.
[4]	Bond R M, Fariss C J, Jones J J, et al. A 61-millionper-son experiment in social influence and political mobiliza-tion[J]. Nature, 2012,489(7415):295-298. doi: 10.1038/nature11421
[5]	刘鲲鹏, 赵宵飒, 胡一睿, 等. 个体及团体异构多方面评分行为建模[J]. 数据与计算发展前沿, 2020,2(2):59-77.
[6]	Asur S, Huberman B A, Szabo G, et al. Trends in social media: Persistence and decay[C]//Proceedings of the Inter-national AAAI Conference on Web and Social Media. 2011,5(1).
[7]	肖宇, 许炜, 夏霖. 一种基于情感倾向分析的网络团体意见领袖识别算法[J]. 计算机科学, 2012,39(2):34-37.
[8]	张昊, 刘功申, 苏波. 一种微博用户影响力的计算方法[J]. 计算机应用与软件, 2015,32(3):41-44.
[9]	Page L, Brin S, Motwani R, et al. The PageRank citation ranking: Bringing order to the web[R]. Stanford InfoLab, 1999.
[10]	Kleinberg J M. Authoritative sources in a hyperlinked environment[J]. Journal of the ACM (JACM), 1999,46(5):604-632. doi: 10.1145/324133.324140
[11]	Lamberti F, Sanna A, Demartini C. A relation-based page rank algorithm for semantic web search engines[J]. IEEE Transactions on Knowledge and Data Engineering, 2008,21(1):123-136. doi: 10.1109/TKDE.2008.113
[12]	Jing Y, Baluja S. Pagerank for product image search[C]//Proceeding of the 17th international conference on World Wide Web. 2008: 307-316.
[13]	Liu Y, Lin Y. Supervised HITS algorithm for MEDLINE citation ranking[C]//2007 IEEE 7th International Sympo-sium on BioInformatics and BioEngineering. IEEE, 2007: 1323-1327.
[14]	Asano Y, Tezuka Y, Nishizeki T. Improvements of HITS algorithms for spam links[M]//Advances in data and web management. Springer, Berlin, Heidelberg, 2007: 479-490.
[15]	王君泽, 王雅蕾, 禹航, 等. 微博客意见领袖识别模型研究[J]. 新闻与传播研究, 2011 (6):81-88.
[16]	Kwak H, Lee C, Park H, et al. What is Twitter, a social network or a news media?[C]//Proceedings of the 19th international conference on World wide web. 2010: 591-600.
[17]	Weng J, Lim E P, Jiang J, et al. Twitterrank: finding topic-sensitive influential twitterers[C]//Proceedings of the third ACM international conference on Web search and data mining. 2010: 261-270.
[18]	石磊, 张聪, 卫琳. 引入活跃指数的微博用户排名机制[J]. 小型微型计算机系统, 2012,33(1):110-114.
[19]	童薇, 陈威, 孟小峰. EDM: 高效的微博事件检测算法[J]. 计算机科学与探索, 2012,6(12):1076-1086.
[20]	李华, 朱荔. 基于影响力的微博新兴热点事件检测[J]. 计算机应用与软件, 2016,33(5):98-101.
[21]	郭跇秀, 吕学强, 李卓. 基于突发词聚类的微博突发事件检测方法[J]. 计算机应用, 2014,34(2):486-490.
[22]	Cha M, Haddadi H, Benevenuto F, et al. Measuring User Influence in Twitter: The Million Follower Fallacy[C]// Proceedings of the 4th International AAAI Conference on Weblogs and Social Media. 2010,4(1).
[23]	杨立公, 朱俭, 汤世平. 文本情感分析综述[J]. 计算机应用, 2013,33(6):1574-1607.
[24]	曹玖新, 陈高君, 吴江林, 等. 基于多维特征分析的社交网络意见领袖挖掘[J]. 电子学报, 2016,44(4):898-905.
[25]	陈涛, 安俊秀. 基于特征融合的微博短文本情感分类研究[J]. 数据与计算发展前沿, 2020,2(6):21-29.
[26]	崔连超. 互联网评论文本情感分析研究[D]. 济南: 山东大学, 2015.
[27]	Bakshi R K, Kaur N, Kaur R, et al. Opinion mining and sentiment analysis[C]//2016 3rd international conference on computing for sustainable global development (INDIA Com). IEEE, 2016: 452-455.
[28]	朱嫣岚, 闵锦, 周雅倩, 等. 基于HowNet的词汇语义倾向计算[J]. 中文信息学报, 2006,20(1):16-22.
[29]	徐琳宏, 林鸿飞, 潘宇, 等. 情感词汇本体的构造[J]. 情报学报, 2008,27(2):180-185.
[30]	徐军, 丁宇新, 王晓龙. 使用机器学习方法进行新闻的情感自动分类[J]. 中文信息学报, 2007,21(6):95-100.
[31]	刘志明, 刘鲁. 微博网络舆情中的意见领袖识别及分析[J]. 系统工程, 2011,29(6):8-16.
[32]	Rogers E M, Cartano D G. Methods of Measuring Opi-nion Leadership[J]. Public Opinion Quarterly, 1962,26(3):435-441. doi: 10.1086/267118

词语	词性	情感	强度	极性
周到	adj	赞扬	5	2
手头紧	idiom	烦闷	7	0

类别	特征
帖子特征	帖子id
	发帖用户id
	发帖用户昵称
	正文文本
	转发/评论/点赞数
用户特征	用户id
	用户昵称
	性别
	阳光信用
	微博数
	粉丝数
	关注数
	微博等级
	会员等级
	认证状态

	核心意见领袖	普通意见领袖	主流人群	少数极端派	P值
总影响力	28.47±5.75	5.66±3.64	0.86±0.86	0.53±0.16	<0.001
平均影响力	0.66±0.08	0.63±0.08	0.52±0.10	0.48±0.11	<0.001
情感极性值	-26.67±19.04	-77.07±86.12	-26.51±101.19	-978.00±258.71	<0.001
立场值	-1788.8±1081.9	-734.4±523.2	-2.3±161.7	-422.8±1165.8	<0.001

用户名	粉丝数	转发数	评论数	点赞数	身份类型	活跃领域
帝吧官微	1104011	50232	46881	631000	网络社区账号	时政,文化
上帝之鹰_5zn	2305345	18143	27344	347728	翻译,自媒体	时政,军事
地瓜熊老六	6372786	13997	13436	122730	作家,自媒体	时政,文化
作者海菱	571272	7338	7148	73706	作家,自媒体	新闻,两性
骑扫帚的老道士	431468	2428	3062	21792	宗教人士,自媒体	新闻,国学