User emotion for modeling retweeting behaviors. (December 2017)

• Record Type: Journal Article
• Authors: Chen, Jinpeng; Liu, Yu; Zou, Ming
• Is Part Of: Neural networks. Volume 96(2017)
• Page Start: 11
• Page End: 21
• Publication Date: 2017-12
• Subjects: Retweeting behavior -- Emotion -- Social network -- Twitter -- Factor analysis; Neural computers -- Periodicals; Neural networks (Computer science) -- Periodicals; Neural networks (Neurobiology) -- Periodicals; Nervous System -- Periodicals; Ordinateurs neuronaux -- Périodiques; Réseaux neuronaux (Informatique) -- Périodiques; Réseaux neuronaux (Neurobiologie) -- Périodiques; Neural computers; Neural networks (Computer science); Neural networks (Neurobiology); Periodicals; 006.32
• DOI: 10.1016/j.neunet.2017.08.006
• Languages: English
• ISSNs: 0893-6080

Abstract

Abstract: Twitter and other microblogs have rapidly become a significant mean of information propagation in today's web. Understanding the main factors that make certain pieces of information spread quickly in these platforms has emerged as a popular topic. Therefore, as a simple yet powerful way of disseminating useful information, retweeting has attracted much interest. Existing methods for retweets have been conducted for analyzing the social network structure, or understanding the retweeting mechanism. However, little attention is paid to whether users' emotion will affect users' retweeting behavior. In this paper, we study the user emotion problem in a large social network. Particularly, we consider users' retweet behaviors and focus on investigating whether users with a certain emotional status will retweet the tweet corresponding with users' current mood from their friends. In order to achieve this goal, we propose a retweeting prediction framework. First, we construct a model of emotion detection via considering two kinds of emotional signals; second, we extract possible retweeted friends and tweets; third, based on the first two steps, we obtain Top-N retweets using Learn-to-Rank method. Experiments are performed on two real-world datasets, the Twitter network and Obama–McCain Debate dataset, with comprehensive measurements. Experimental results demonstrate that our retweeting prediction framework has substantial advantages over commonly used retweeting prediction approaches in predicting retweeting behaviors. Consider Precision in Twitter network as an example. For the Top-N stage, our method can, on average, increase by 15.2% and 11.2% in relation to Tweet(+SV) and User(+ED), respectively. We find that emotion is a vital feature which affects retweetability.