Share this article:
Tweeting Seductively: Maximizing the Social Influence of Twitter Messages
Most of us are familiar with Twitter. Users post and respond to short messages (tweets) on the platform, conveying news and opinions on various topics, including politics, entertainment and business.
With hundreds of millions of tweets posted every day, the network potentially impacts the thoughts, feelings, attitudes and behaviours of over 300 million network users worldwide. With every tweet capable of flitting across such a vast network, many researchers are studying the social influence of tweets. Just what is it that makes people sit up and respond to some tweets, whereas others simply glide on by, barely noticed?
One such researcher is Siwar Jendoubi of the LARODEC laboratory at the University of Tunis in Tunisia. In a paper just published in Elsevier’s journal Knowledge-Based Systems, she and her colleagues introduce a method to measure influence, and then propose and test two models that aim to maximise this influence.
Let’s say a company starts to market a new product. One promising way to do this is to select a set of influential users in the Twitter network, and target them with gifts or discounts to entice them to take up the product. The selected users might then encourage their friends to buy the product too. And then friends of their friends, until by ‘word of mouth’ the product achieves substantial sales. But for this strategy to be successful, it is important that the company target the right users in the first place. So how exactly does it do that?
Jendoubi and colleagues believe the answer lies in first identifying various significant factors, such as the importance of the user in the network structure and the popularity of user’s tweets. They then employed the theory of belief functions, or Dempster-Shafer theory, which provides an algorithm to calculate a concrete measure of influence. “The paper introduces a new influence maximisation approach that allows us to detect the best set of influencers,” Jendoubi explains. “Those influencers can trigger the largest cascade of ad propagation through the network, which allows the company to maximise its profits with the lowest marketing cost.”
The researchers then applied this approach to compare their two proposed models of influence maximisation with existing models. They found their models are significantly better at selecting influencer users in terms of activating Twitter features, namely Follow, Mention, Retweet and Tweet. “Compared to existing approaches,” Jendoubi concludes, “the proposed solution detects a set of influencers having the best quality in terms of connectivity, activity and propagation.”
In future research, she hopes to further develop this approach, making the results more precise and applying it to other social networks like Facebook and LinkedIn.
Read the article free online via ScienceDirect. Jendoubi, S.: “Two evidential data based models for influence maximization in Twitter,” Knowledge-Based Systems (2017)
If you found this article interesting, you may also wish view more books and journals on similar subjects. For more information
The general scope of social sciences is vast, and Elsevier’s collection of journals, books, and eBooks examine in detail a wide range of topics in this area, from sociology, law, and cognitive science to political science, education, and linguistics. Our Chandos imprint in particular, known for high-quality scholarship in Asian studies, library and information science, and business management, reflects Elsevier’s continuing commitment to these crucial areas of study.