Article: Analysing user retweeting behaviour on microblogs: prediction model and influencing features Journal: International Journal of Computational Science and Engineering (IJCSE) 2017 Vol.15 No.3/4 pp.176 - 187 Abstract: This paper explores the feasibility of predicting users' retweeting behaviour and ranks the influencing features affecting that behaviour. The four first-dimension features, namely author, text, recipient and relationship are extracted and split into 39 second-dimension features. This study then applies support vector machine (SVM) to build the prediction model. Data samples extracted from Sina Microblog platform are subsequently used to evaluate this prediction model and rank the 39 second-dimension features. The results show the recall rate of this model is 58.67%, the precision rate is 82.19%, and the F1 test value is 68.46%, which show that the performance of the prediction model is highly satisfactory. Moreover, results of ranking indicate four features affect retweeting behaviour of users: the active degree of microblog author, the similarity of interests between the author and the recipient, the active degree of microblog recipient and the similarity between the theme of microblog and the recipient's interest. Inderscience Publishers - linking academia, business and industry through research

Title: Analysing user retweeting behaviour on microblogs: prediction model and influencing features

Authors: Chenglong Lin; Yanyan Li; Ting-Wen Chang; Kinshuk

Addresses: School of Educational Technology, Beijing Normal University, Beijing, 100875, China ' Smart Learning Institute, School of Educational Technology, Beijing Normal University, Beijing, 100875, China ' Smart Learning Institute, School of Educational Technology, Beijing Normal University, Beijing, 100875, China ' School of Computing and Information System, Athabasca University, Edmonton, AB, T5X 2T9, Canada

Abstract: This paper explores the feasibility of predicting users' retweeting behaviour and ranks the influencing features affecting that behaviour. The four first-dimension features, namely author, text, recipient and relationship are extracted and split into 39 second-dimension features. This study then applies support vector machine (SVM) to build the prediction model. Data samples extracted from Sina Microblog platform are subsequently used to evaluate this prediction model and rank the 39 second-dimension features. The results show the recall rate of this model is 58.67%, the precision rate is 82.19%, and the F1 test value is 68.46%, which show that the performance of the prediction model is highly satisfactory. Moreover, results of ranking indicate four features affect retweeting behaviour of users: the active degree of microblog author, the similarity of interests between the author and the recipient, the active degree of microblog recipient and the similarity between the theme of microblog and the recipient's interest.

Keywords: microblog; retweeting behaviour; prediction model; influence ranking; support vector machine; SVM; information gain.

DOI: 10.1504/IJCSE.2017.087405

International Journal of Computational Science and Engineering, 2017 Vol.15 No.3/4, pp.176 - 187

Received: 14 Apr 2016
Accepted: 27 Aug 2016
Published online: 15 Oct 2017 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article

Title: Analysing user retweeting behaviour on microblogs: prediction model and influencing features

Keep up-to-date