Twitter: A key player in the 2011 World Wide Web conference

The 2011 World Wide Web conference took place in Hyderabad, India from March 28th through April 1st. More than 800 researchers from around the world gathered to learn about the latest research related to the Web. The most important part of the conference was the research papers, 90 of them were presented in the last three days.

Twitter data played a major role in many of the research papers. To get an idea, four out of the 81 papers had the word Twitter in the title and 21 in the text. The main reason for such a large number of papers related to Twitter is probable that Twitter data relatively easy to obtain. However, there are other reasons that make Twitter interesting to study.

Information is labeled. Hashtags have become increasingly popular on Twitter. Users label their tweets with hashtags, making it easier for researchers to automatically detect information diffusion, get an idea of what a tweet may be about without reading the whole text, and identify trends. For example, there is a paper that makes use of hashtags to explore the differences in diffusion for various different topics on Twitter.

The microblogging site has a mixture of social and informational relationships. Many of the relationships on Twitter are truly social, that is, people who actually know each other follow each other. At the same time, many relationships are purely informational. Indeed, many people use Twitter as a tool for obtaining the information they are interested in by following their favorite news sources, celebrities, journalists and bloggers. This mixture of following relations creates a notion of status or class in the network. On the one hand, we have the users that are followed because of social reasons, and on the other, we have those who are followed for their information.

This is a great opportunity to study the interplay between what friends say to each other and what more prolific groups say to the rest of us. For example, there is a paper about the how different groups of people contribute differently to Twitter depending of their status.

Furthermore, this mixture of types of users creates a credibility problem in the network. Users are faced with the question of whom they should trust and whom they should not. This is a basis for interesting research. For example, a paper studied the features that make a user’s tweet be perceived as trustworthy.

In general, online social networks make it possible for us to learn about how people communicate at a large scale. Many of the previous findings about human communication patterns were based experiments with a very small set of people. With Twitter, we can study millions of people all at once. For example, it has been known that we tend to coordinate in many ways when we talk to each other. A paper showed that these trends also hold at a large scale on Twitter.

Finally, when we obtain data from a social network, we often find that it is not complete. An interesting challenge is to come up with creative ways to deal with this by approximating what it is missing. Since Twitter does not tell us when a following relation started, a paper showed a way of approximating these times.

While some people in the community were, perhaps, overwhelmed by so many so called “Twitter papers”, the reasons for using Twitter data are very powerful. More importantly, one can use Twitter data sets to study many different research topics. Some people may still lump all “Twitter papers” into a single research category without realizing that there is much more to it. Twitter is only a tool to study more general phenomena.