“But who is going to read 12,000 tweets?!” How researchers can collect and share relevant social media content at conferences

Using social media to communicate at conferences allows more space and time for attendees to join discussions and network. But are these conversational tweets of any use once the drinks reception begins? Nicole Beale and Lisa Harris hope that archiving and visualizing the data will spark new research routes.

In our previous post, we introduced the SMiLE project and the ways that we had implemented the use of social media at the CAA2012 conference to test the extent to which we could use the tools to support the following:

• Community building
• Administrative support
• Networking between individuals
• Resource discovery

Since the conference, we have recruited a number of MSc students who are drawing upon the social media data from the conference to write dissertations over the summer. Their topics include bridging the ‘tweeting divide’, managing online communities in the context of live events and the opportunities and challenges posed to researchers by data visualisations. When completed, we plan to publish summaries of all these project findings as freely available downloadable resources. We are also working with the Oxforde-ResearchCentre to develop a code of conduct and best practice guide for the collecting, curating and archiving of social media data based on our experiences so far.  

The CAA2012 conference at which the SMiLE project began included a number of archaeologists interested in data preservation and publication, for example from the ArchaeologyDataService (ADS) at the University of York. We work closely with the ADS already at Southampton and we are now discussing deposit of the SMiLE archive with them, and writing a publication for the InternetArchaeology journal that will explicitly link between the underlying data (the tweet archive) and a journal article exploring the archive and its creation. Such a linked archive has as its origin theMakingtheLEAP projects developed by the ADS and Internet Archaeology.

Discussion of the issues involved in depositing a tweet archive began at the conference and are included in outline form inthisStorify. The ADS are interested in archiving the social media content but we have identified three key issues.

Firstly, the ADS repository is immutable – data deposited in any archive are intended to remain intact forever. The legal position is clear – tweets are published in the public domain and hence are freely available for use by anyone. But this is at odds with any ethical choice on our part that presupposes the ability for people to delete tweets from the archive at any stage. Secondly, deposit with the ADS requires agreement from those who created the data. Normally this is via formal sign-off but given the nature and scale of the tweet archive such a process would be impracticable. Thirdly, the decision to archive the CAA social media data with the ADS was made after the event had happened, and so a framework for the archive submission was not agreed beforehand.  Had we anticipated the usefulness of the archive early on in our project, the ADS would have been able to advise the team on the most appropriate actions for both gaining permissions to archive and the selection policy. We are now working with them to design such a framework.

So, we are left with some big decisions to make. Last year a paper (PDF) by Kaitlin Costello and Jason Priem in the Journal of the American Society for Information Science described their research which sought the opinions of twitter users about the archiving of tweets. The results were quite negative – although most people interviewed said that tweets should be archived, particular concerns were expressed about 1) institutional archiving (as being analogous to the recording of phone calls by one’s boss) and 2) the possibility of individual comments being taken out of context and used against the author in the future. The authors noted that the decision by the US Library of Congress to archive tweets (which took place during their data collection) had a notable positive impact on the acceptability of tweet archiving to their interviewees.

More recently, and as we have described above, few individuals in the CAA community expressed concern about our plans to archive conference tweets. The issue is not just saving the data, but preserving it in a way that is meaningful and useful to researchers.

Particularly thought-provoking were those comments that considered how real value could be added to the 12,000 tweets available online.

One survey respondent said,“Who’s going to read all those 12,000 messages?” And another: “Basically there is no use saving it all. Making informed selections and processing it into a desirable and accessible format would be best.” Specific suggestions for making the data more useful included linking specific tweets to papers as they were presented, and also incorporating later tweets and feedback relating to individual papers.

As part of the preparations for the depositing of the archive with the ADS, all users who had used the #caasoton hashtag were contacted and asked to complete an opt-outformfor any tweets that they wished to have removed from the twitter archive however no users came forward.

Separately, within the post-conference feedback process, all delegates were asked to comment on what they thought the future of the twitter archive should be. 151 responded to the survey. The majority of respondents preferred that the archive be kept, with most preferring submission to the ADS.  Out of those participants who wished for the survey to be submitted to the ADS, over half wanted twitter users to be given an option to opt-out of the archive. The pie chart below gives an overview of the results:

The challenge now is to work on designing interfaces that allow users to investigate the data in the most useful way.  Mark Borkum (@markborkum) has provided some initial visualisations of the Twitter data, and Tom Brughmans (@tombrughmans) is working on looking at the data in more depth.  The visualisations use network analysis to illustrate the relationships that exist between Twitter users, through linking different information together, such as shared hashtags, or retweets.  The pictures below show the network of the CAA2012 tweets. Tweets are represented by the twitter bird, and hashtags by folder icons, and each line links the tweets to the hashtags that are contained within it.  In the first picture, the #caasoton hashtag is included in the visualisation, and the image shows a tightly connected network, with a centralised theme of conversation. 

When we remove the #caasoton hashtag, a few things are apparent.  Firstly, other clusters of common topics of conversation are happening that include the #caasoton hashtag. 

On closer examination, these are general hashtags that users have made up that cover more specific things happening at the conference, such as session-specific hashtags, like #caa_theory10 which has topic specific hashtags such as #datagovuk and #opendata connected to it in the diagram.  Secondly, whilst there are smaller conversations going on between one or two users, most conversations are occurring between multiple users, with quite a lot of activity around each hashtag.

Tom Brughmans, a SMiLE team member who specialises in network analysis, has been creating beautiful visualisations of the Twitter archive, and we will be making more of these available through the CAA website.

Our Next Steps
The team plans to create visualisations of the archive that will provide alternative ways to explore the data, such as timeline based interfaces and interactive search applications.

In particular, we will be investigating how the use of social media post-event can facilitate discussion around new research, and in building ties between institutions and individuals working in the area of computational archaeology. Please let us know if you would be interested in making a specific contribution to this. 

Note: This article gives the views of the author(s), and not the position of the Impact of Social Sciences blog, nor of the London School of Economics.