Exploring the Twitterverse with Vista

By Everett Dorma Posted: July 3, 2015 6:00 a.m.

The Vista research team (L-R) Khantil Patel, Dr. Orland Hoeber, Maha El Meseery, Kenneth Odoh, (missing Radhika Gopi).
The Vista research team (L-R) Khantil Patel, Dr. Orland Hoeber, Maha El Meseery, Kenneth Odoh, (missing Radhika Gopi). Photo: U of R Photography

Computer Science Professor Orland Hoeber and his graduate students are using Twitter posts about Le Tour de France to help develop Vista (Visual Twitter Analytics), a computer program with a new interactive approach to analyzing Twitter data.

“Twitter is a window into public opinion about a wide range of events and topics,” says Dr. Hoeber. “Vista has been designed to explore and analyze tweets to obtain public opinion about products, companies, and industries, which is useful for marketing, risk and brand management, financial investment and many other sectors interested in understanding public opinion.”

As sport fans identify very strongly with their favourite sport, team, and athletes, what they post to Twitter can provide valuable insight not only in the direct context of sport, but also into human behaviour and underlying attitudes.

Tweets about Le Tour de France also provide a means for testing Vista in the context of big data. To date, Hoeber’s team has collected over 400,000 tweets from the 2013 Le Tour de France, another 500,000 from the 2014 event, and expect to be close to 700,000 for the 2015 event this summer.

“Analyzing this large amount of data using manual methods would be incredibly time consuming and using sampling methods can miss important tweets and interactions between the tweets,” says Hoeber.

What sets Vista apart from the commercial software systems that exist for analyzing Twitter data is the highly interactive features provided to support the analysis and exploration of the data.

While many different systems will show a timeline and map, Vista supports the dynamic and coordinated filtering of each. In addition, hashtags, terms, user mentions, and authors can be used to first identify the common themes in the data, and then generate new timelines that filter out uninteresting features or focus on interesting aspects that have emerged from the data.

As a result, Vista doesn’t just show you what people are saying on Twitter; it allows you to dig into the data to discover and analyze interesting sub-topics that would be difficult to find otherwise.

The initial development work for this project was funded by a Start-Up Grant provided by the Faculty of Science, and New Faculty Matching Funds provided by the Vice-President Research. Subsequent work to extend Vista and explore how it may be used to support investment analytics has been funded by Greystone Managed Investments Inc. and Natural Sciences and Engineering Research Council's (NSERC) Engage and Engage Plus grants.

Research impact is an area the University continues to focus on as one of the three priority areas in its new strategic plan.