IBM Research: SXSW Spotlight: Data as Narrative

Editor’s note: This brief Q&A series will feature IBM researchers making presentations at the 2012 South-by-Southwest Interactive Conference in Austin, Texas.

Join the conversation: #sxswibm #timeMap

Research Scientist Dr. Jennifer Thom, a member of IBM’s Visual Communications Lab, is part of the Maps of Time: Data As Narrative panel – a discussion about understanding and visualizing data over time – on Monday, March 12.

Q: What are the kinds of thing you work on as a researcher at IBM's Visual Communications Lab?

I study online social behavior and focus on how social media can influence and create new opportunities for collaboration. I’m especially interested in how these tools can help distributed groups – who speak different languages and work in different time zones – work together more effectively.

Q: What's an example of a data visualization project you've worked on?

Social media is incredibly multilingual, and the systems within IBM are no exception. Many IBMers contribute to our social tools in many native languages, which is helpful for creating community for a global enterprise. At the same time, there are multilingual IBMers who have varying levels of fluency between the different languages they speak, yet they encounter and consume content in their non-native languages in their daily work.

I was interested in improving the reading experience for multilingual IBMers – especially since an increasing amount of information is shared in these spaces as our business becomes more social.

In collaboration with a graduate student at MIT, we’ve worked on visually transforming online blogs to improve the reading experience for IBMers who consume content in their non-native language. From user data that we gathered, we developed a set of design criteria to reduce visual distractions in order to improve scanning and skimming of social software content.

Initial evaluations of this approach are promising, as global IBMers have indicated that this approach has made the reading experience of this content more enjoyable.

Q: At this year's SXSW, you're on a panel about "Data as a narrative." How does this idea go beyond something like Facebook's timeline?

I’m interested in using the past to help predict the future and looking at how we can learn from the data that we share on these social media systems to solve problems.

For instance, one recent project that I completed looked at the #stuffibmerssay meme that emerged on Twitter at the end of 2011. A small number of IBMers spontaneously created this hashtag to append to humorous tweets that commented on different aspects of life as an IBMer.

The number quickly grew over a period of few days, where different IBMers contributed their perspectives and experiences and created a shared experience for IBMers who often work in diverse environments around the globe. When we looked at the content of the tweets more closely, we realized that collectively they helped expose aspects of our organizational culture.

Since then, we’ve been thinking about ways that we can use these tweets as a barometer of sorts: whether about IBMers, or the systems we maintain, develop and deliver. We’ve also thought about using memes as an elicitation tool to get a sense of what people are working on or thinking about.

*Note: this research will be presented at ICWSM 2012 in Dublin in June.

Q: How might the idea of cobbling together all of a person's (or company's) online data look? Would this have to be a new social network?

So, I think the project that I just described is one approach in where we can leverage people’s existing behavior. The challenge is in aggregating multiple feeds over multiple systems and helping users make sense of the fire hose.

One approach would be to create better visualizations of this data so that people can make sense of what’s being put out there.

Q: How could a person -- or a business -- use a timeline of their online lives, while maintaining security and privacy?

That’s a huge challenge since the same aggregation can help us become more predictive can uncover patterns.

I suspect the best answer to this is a combination of policy and design – helping individuals better opt-out of this aggregation, and trying to figure out what the right amount of noise so that the aggregation of data and timelines doesn’t make individuals so immediately identifiable.

Q: What's a favorite data visualization of yours? What makes it a good example?

I recently came across Yanni Loukissas’s visualization of the Apollo 11 moon landing and thought it was a great example of storytelling through disparate types of data linked by time.

In particular, I thought the combination of the audio channel with the output from the different computer systems involved in managing the launch was a nice integration of the social and technical aspects of complex coordination.