Exploring RSA16 Twitter Data
The Coalition’s Director of Digital Media and Outreach Patricia Fancher asked me to write up some reflections about the data visualizations that I created with a corpus of tweet data from the 2016 Rhetoric Society of America conference. I took up the offer, because I wanted to get to know the Coalition more and also take up this opportunity to reflect a bit more about the tweet data.
If you need some context, I created some chord diagrams (see Fig. 1) that represent some isolated relationships between particular keywords used by people who tweeted about different sessions during the conference. I shared my work on the RSA Facebook page, and Patricia also shared it on the CRSHRC Facebook page.
Feel free to interact with the diagram on my website in a new tab, then come back to read on.
Generally, chord diagrams represent inter-relationships between different datum in a matrix. In this case, I developed a matrix of keywords and the number of times the keywords are mentioned together in a tweet. In the diagram, keywords are represented by the arcs that makeup the radial part of the circle. The length of the arc represents the total number of cross-mentions of each keyword. The chords linking different arcs represent the number of times the keywords are mentioned together in a tweet. Also an important note: any asterisk by a keyword means that I used a regular expression pattern to consolidate tense and closely related keywords. (See this example regex pattern for embodiment.)
Patricia suggested some of the following questions for me to consider:
- “How may twitter reveal the different coalitions of feminists or coalitions of rhetoricians, where we identify coalitions through grouping/chords?
- (How) Is tweeting a feminist practice? Does any of this material help to recommend the practice of conference tweeting to feminists/academics?
Regarding the first question, Twitter and conference tweeting have a lot of layered complexity that I can’t account for with my data work here. In my opinion, tweeting during a conference doesn’t automatically help me to develop clear connections for and between different coalitions and colleagues — not without substantial work to seek out and forge those connections (cf. Patricia’s work with the Coalition and how Women in Technical Communication uses the #womenintc hashtag). Again, my opinion about tweeting and coalition work is not grounded in my data work here. However, I think if conference organizers created a team of people who planned, collected, processed, and analyzed social media data, they could help colleagues carry out coalition-building and invent new feminist social media practices.
Inventing new feminist practices with social media data could become a more integral part to conference planning, which touches on Patricia’s second question. For instance, there seem to be connections between keywords related to race, latin/x, and disability (see Figures 2 and 3). The connections between race and latin/x make some immediate sense to me, but the links between those 2 and disability seem worthy of further investigation. With more planning and support upfront, I could have processed the data differently to analyze interesting relationships such as these. At the very least, I would have been able to supplement the chord diagram with a list of the tweets and their tweeters for each chord. Conference organizers, coalitions such as CRSHRC, and scholars could use this information to connect people and developing research.
A more robust social media strategy could also help scholars consider trends surround particular kinds of research and scholarly domains. For example, the link between Archive* and Digital* is strong (see Fig. 4), which makes me wonder: What topics and problem areas are scholars concerned about between these two keywords. What is being archived digitally? And by whom? Rhetoricians invested in this research domain could start mapping trends and scholars, and analyze the tweets related to such work. The same holds true for strong links between Method* and Gender, Race*, and Feminist* (see Fig. 5). By providing this type of information, teachers could start asking their seminar students to conduct analysis of these trends as a complimentary method to traditional literature reviews.
Another social media strategy could include identifying a lack of overlap between keywords and research domains, which perhaps deserve more attention. After reviewing the chord diagram, I find it interesting that there was no or little overlap between Race and Disability and the following keywords: Object, Digital, Material, and Machine. Mapping and archiving weak connections is just as important as those with recurrently strong cross-mentions.
Future considerations about conferences and Twitter Data
By reflecting on this diagram, I hope to provide some new avenues to explore, take-up, curate, and guide coalitions and colleagues with data from conference social media practices. Much of these ideas, visions, and questions could become a reality if conference organizers created and released basic data points that could be combined with social media data sets: panels with panel titles, abstracts, keywords, session labels, times, etc. As a researcher, if I had this data and setup a number of people from across the discipline to tweet all of the panels, then I could conduct an analysis across time and against other panels at the time. I could also potentially explore ratios related to keywords, or even how the panelists’ keywords relate to the keywords used to describe the panel. Social media researchers could also conduct feminist studies of representation of conference panels and/or how panels are represented in and through social media practices. Overall, conference organizers could potentially create and facilitate numerous new scholarly functions, if they released panelist-defined data in a usable data and file format.
What other ways could conference organizers utilize and organize people who tweet to help facilitate better tweet data? If a conference created and released this panelist-defined data in a usable format, I see potential field-building opportunities too. Session times could be mapped onto people who attended and tweeted the panels. Panelists, or whomever interested after-the-fact, could then review those tweets and scholars who wrote them to make new connections and encourage future participation.
These are just a few ideas that I have been mulling over since working with this Twitter data. What questions and comments do you have about social media data and conference organization? Feel free to share your ideas on the Coalition’s Facebook page or tweet at me, @lndgrn, and use the hashtag #confdata on Twitter.