One of the things that has set Twitter apart in its short but fast development history is how the users have adapted Twitter for use in ways which were never intended by its developers. People use Twitter in many different scenarios and of the common and prevalent usage is in conference settings.
This was the problem which we studied in the Personal Information Management class at Virginia Tech in the Spring semester of 2010. I joined hands with two other passionate Twitter users Edgardo Vega (@CasaDeVega) and Josette Torres(@girlinblack) from the English department to study how people are using Twitter at conferences. Fortunately, our professor Dr. Manuel Perez Quinones (@mapq) is also an ardent Twitter user and this helped a lot!
Twitter is a huge personal information repository and there were many motivations to study this problem. This is what we did
- Randomly selected twenty users who attended SXSW 2010 based on the Klout influence score.
- Using the Twitter API, we scrapped all the tweets made by the users for a five week time period. We obtained a little over 10000 tweets for the time frame which included two weeks before the conference, the conference week and two weeks after the conference.
- Performed quantitative analysis measuring the number of tweets/mentions/replies etc
- Performed qualitative analysis by manually assigning all the tweets into 15 different categories. Some of the categories were ‘location’, ‘informational’, ‘observation’, ‘emotion’ etc
- Inferred results from the data analysis.
PS: There are many big holes in our analysis methodology and the results are not perfect. We planned to do a lot on the data part but due to time constrains we weren’t able to. Nevertheless, this was a very good experience and definitely one of the different projects that i worked on in grad school.
I have embedded the final paper below and the download link for the PDF version.
