Patterns of Implicit and Non-follower Retweet Propagation: Investigating the Role of Applications and Hashtags

Existing literature on retweets seems to focus mainly on retweets created using explicit, formal retweeting mechanisms, such as Twitter's own native retweet function, and the prefixing of the terms 'RT' or 'via' in front of copied tweets. However, retweets can also be made using implicit, informal mechanisms. These include tweet replies and other mechanisms, which use neither the native nor RT/via mechanisms, but their content and timelines suggest the likelihood of being a retweet. Moreover, retweets can also occur with or without a defined follower/following network path between a tweet originator and a retweeter. This paper presents an initial taxonomy of propagation based on seven different ways a tweet may spread: native, native non-follower, RT/Via, RT/Via non-follower, replies, non-follower replies and other implicit 'retweets'. An experiment has examined this new model, by investigating where tweets containing URLs from the domains of online petitions, charity fundraisers, news portals, and YouTube videos can be classified into the seven different categories. When including other implicit 'retweets', more than 50% of all the retweets found across all four domains were classified as implicit retweets, while more than 79% of all retweets were made by non-followers. More work needs to be done on the composition of other implicit 'retweets'. Initial investigations found hashtags in 99-100% of these tweets, suggesting that retweeting using conventional mechanisms may not be the main method that URLs get propagated across microblogs.