Q. Dhiraj – thanks for taking the time to speak to us today. Even though your Lab is our newest WSTNet member, you’ve been analysing media for long time. Can you talk to us about your journey into social media and Web Science?

A. Of Course. My interest started doing post-graduate research in Sociology at Cambridge where I was looking at how traditional (non-technical) social spaces and interactions were being supplemented, augmented and even replaced by what (at the time) were new technologies such as blogs, newsgroups and forums (all async, pre-web technologies). I was particularly interested in how group identities were affected by the technologies that were mediating the interactions. I looked at the way in which musicians collaborated using technology vs. live interactions and how this might impact participants’ sense of identity, say, in terms of race/ethnicity. My work went on to look at impacts and opportunities around natural disasters such as hurricanes and we’ve gone on to study several US as well as international incidents. All of this work required the collection and analysis of what we could broadly call (in todays terms) social network data, which could be classified, mapped and visualised to help uncover and understand the observed effects. That process has been going on for me since 2006.

Q. Even those early projects and research questions still sound very relevant and quite contemporary. Would you say that during that time the process of developing research questions has remained relatively stable whilst the types and volumes of data that can be employed have changed rather more substantially? 

A. I witnessed a great deal of competitive system development for social messaging systems working internationally during the dot-com era and the key changes seemed to the ubiquity of messaging standards like the SMS text message (a 140 character format that Twitter later adopted) based on the growth of mobile networks and, critically, the rise of social “platforms” like Twitter (and later WhatsApp) which transformed the culture of what had been a private point-to-point messaging model into high-speed, real-time shared messaging spaces with API’s that (initially) disclosed information (both data and metadata) about the networks of content and networks of users across multiple locations. Twitter was instrumental in developing this model into what became a new opportunity for disciplines like Web Science to do detailed analysis on huge data sets.

Q. So if the availability of data and data types have driven/enabled the research in this way what has been your experience with the transition of Twitter to X and the loss of access to the Twitter API? 

A. The broader issues with social media APIs, data scraping bans and the resulting legal battles have obviously shaped the way in which data can be gathered/analysed and, arguably, is transforming (has transformed) what it means to do Web Science; but equally we have seen a continuing trend in which the Web/Internet overall has become less overtly text-based and much more visual with the enormous growth in video platforms. This means that as Web Scientists we have had to innovate and develop new/better techniques around computer vision, video analysis and the currently available data sets to do quality research. We now combine our data archives with new data, new ways to annotate and analyse data using mixed methods to be able to work with “small data” at a more personal level vs the level of firehose (i.e., complete) data sets that are not currently (no longer) available.

Q. If you are looking at more video data will the recent rise of high-quality (deep fake) AI generated video cause you particular difficulties?

A. Well bots and fake data have been around (in a smaller way) since the very beginning – there were simple bots to be found in early news groups so fake data and bots are not a new thing at all – though the scale and sophistication of the most recent examples is obviously more concerning and hence we are also looking at how we might better detect bad data and misinformation. 

Q.. Is that your main area of interest?

A. Not only that. We continue to look at ways in which the Web may (dis)empower society and how we might identify and promote (or inoculate against) those effects. We continue to look at group social behaviours during natural disasters where we have followed a number of US and international hurricane events. We’ve studied how cancer is reported on Twitter and how this relates to disease incidence across regions/groups as well the enculturation of young people into vaping (i.e., e-cigarettes) and how much impact social media images and messages may have in that process. But we have also been looking at identifying misinformation and tools (beyond labelling) to help users identify misleading information and how it spreads. 

Q. Many thanks for spending time to talk about your work and the Computational Media Lab. We have listed some papers and a link to your website below.


Dhiraj Murthy is the head of the WSTNet Computational Media Lab at the University of Texas at Austin.

To read more about Dhiraj’s work and the Austin Lab click below