Naaman, Mor, Jeffrey Boase, Chih-Hui Lai. “Is it really about me? Message Content in Social Awareness Streams.” ACM Digital Library, ACM, dl.acm.org/citation.cfm?id=1718953. Accessed 29 Aug. 2017.
Akshay Java, Xiaodan Song, Tim Finin, Belle Tseng. “Why We Twitter: Understanding Microblogging Usage and Communities”. http://aisl.umbc.edu/resources/369.pdf. Accessed 29 Aug. 2017.
Summary
“Is It Really About Me?” is an analytical research paper attempting to determine classifications of Twitter users. The classifications found were “Informers”, users who share a lot of non-personal information, and “Meformers”, users who share a lot of personal information. The researchers were able to cluster users into these groups with strong statistical significance. They did this by first breaking down user posts into different categories (Information Sharing, Me Now, Opinions / Complaints, Random Thoughts, etc.) and then clustering users based on number of occurrences of those categories of posts.
“Why We Twitter” is another analytical research paper with a broader goal of observing some topological and geographical aspects of Twitter as an example of microblogging. They showed special interest in the content shared by different communities and the inter/intra-community connections and patterns. They found that North America (followed by other industrialized continents) is responsible for most Twitter activity. The also found that the content of posts changed through the week, with “school” and “work” being drowned out by “party” and “friends” approaching and during the weekend. The researchers found categories of user posts including: Daily Chatter, Conversations, Sharing Information, Reporting News. These resulted in classifications of users including: Information Source, Friends, and Information Seeker.
Reflection
It is interesting to note the similarities in analysis results and interpretation between the two studies, despite the difference and intentions and methods. Specifically, the categorization of posts leading to the classification of users. Both studies ended up picking out a similar “informer” class; “friends” and “meformers” seem to have largely the same posting habits; “Why We Twitter” put special emphasis on users who did not create much if any original content. It is fascinating to look at these studies and the patterns that they detected. Twitter is a great source of digital data produced by the living organisms that it is about.
Questions
- How else could user posts be categorized?
- Does the intention of a study have a strong effect on the interpretation of the data?
- What would the results of blind machine learning look like if it were also to categorize posts and classify users?
