Labeled LDA: A supervised topic model for credit attribution in multi-labeled corpora
Today, it is the norm for online social (OSN) users to have accounts on multiple services. For example, a recent study showed that 34% of all Twitter users also use Pinterest. This situation leads to interesting questions such as: Are the activities that users perform on each site disjoint? Alternatively, if users perform the same actions on multiple sites, where does the information originate? Given the interlinking between social networks, failure to understand activity across multiple sites may obfuscate the true information dissemination dynamics of the social web. In this study, we take the first steps towards a more complete understanding of user behavior across multiple OSNs. We collect a sample of over 30,000 users that have accounts on both Twitter and Pinterest, crawling their profile information and activity on a daily basis for a period of almost three months. We develop a novel methodology for comparing activity across these two sites. We find that the global patterns of use across the two sites differ significantly, and that users tend to post items to Pinterest before posting them on Twitter. Our findings can inform the understanding of the behavior of users on individual sites, as well as the dynamics of sharing across the social web.