IBM taps social media at Australian Open

Twitter serves feedback on event and informs IBM when to increase capacity.

Lleyton Hewitt's performance last night was measured and analysed by IBM systems. Credit: Adam Bender.

Lleyton Hewitt's performance last night was measured and analysed by IBM systems. Credit: Adam Bender.

IBM and Tennis Australia, partnering for the 20th year at the Australian Open, are analysing social media not only to measure players’ popularity but also to ensure there is adequate capacity to handle spikes in user activity.

See a Computerworld Australia behind-the-data slideshow of the Australian Open 2013.

IBM is analysing the “sentiment” of Twitter comments about players at the Open, according to IBM's sponsorship strategy lead, Elizabeth O’Brien.

By looking at keywords, IBM can automatically determine whether a tweet is positive or negative, she said. It can then compile the data to rank players in a social leaderboard on the Australian Open website.

The leaderboard is updated every few minutes. After the event, IBM compiles an index providing feedback on the success of the event. Tennis Australia makes use of the feedback for marketing purposes, said Tennis Australia CIO Samir Mahir.

“It’s more complicated than you’d think to figure out what’s a positive sentiment and what’s a negative sentiment,” O’Brien said. For example, the word “bad” is sometimes used in slang to mean “good.”

An IBM official said context can be determined by looking at a combination of keywords in a given tweet. The system also refers to a list of slang and curse words, another official said.

Social media is also a key input into a predictive cloud provisioning technology debuting at this year’s Open, O’Brien said. This “intelligent” cloud system combines social media awareness with the schedule and historical data to predict spikes in activity, automatically increasing computing power before a problem can arise.

To gauge the level of social media activity, the system scans Twitter, Facebook, YouTube and blogs, O’Brien said.

“It’s combining analytics and the cloud,” she said. “It’s the system actually recognising” a pattern and automatically scaling up capacity.

In addition, Tennis Australia makes use of social media in a mobile ticketing app developed with Ticketek, Mahir said. The app lets customers locate their friends in the venue.

Keeping Score

Tennis Australia uses IBM infrastructure to input score data and automatically distribute it to a variety of destinations, including scoreboards, on-site media and the Australian Open website and official smartphone app.

“We have to engage our fans whether they’re here on site or viewing on TV or online,” Mahir said.

It starts with the chair umpire entering each play into a “Chump”, a touch-screen device with a similar size and appearance to a retail payment terminal.

A look at the Chump, used by chair umpires to enter each play.
A look at the Chump, used by chair umpires to enter each play.

The Chump is hardwired into the IBM network, and immediately sends the data to all the destinations. Tennis Australia believes it can get better reliability from a “good old wired network” compared to a wireless system, said Mahir.

Tennis Australia has 200TB to store all the data and video, which it expects to be enough for two years, Mahir said. “Last year, we used about 50TB.”

One destination for data distribution is the media room, where IBM has installed about 300 touchscreen devices for journalists. The devices display information including statistics from each ongoing match, match analyses and live and recorded IPTV video.

The media room at the Australian Open.
The media room at the Australian Open.

Match analysis is also provided to players and coaches after the contest. This is the first year match analysis is available online, Mahir said. It was previously available only on DVD and over IPTV, he said.

The top Web features demanded by fans are scores, results and video, Mahir said. However, data sent to the website provides more than scores. One tool charts a player’s momentum, highlighting turning points in each match. Another feature called Slam Tracker considers player types and historical statistics to predict keys to winning each match.

For each match, Slam Tracker assigns each player three performance goals they need to complete to make victory likely. If one player achieves all three of their keys and the other players gets none of theirs, the system is correct more than 95 per cent of the time, O’Brien said.

The predictive tool was introduced last year but IBM continues to make enhancements to improve the accuracy of predictions, O’Brien said. Tennis Australia uses a similar tool year-round to help coaches, added Mahir.

Scores and live video are also available through an IBM smartphone app available for Apple iOS and Android. More complex features on the website like the keys to winning the game are not available on the app.

More mobile devices are being used on site than ever before, Mahir told Computerworld Australia. He said that 36 per cent of all website visits were accessed on a mobile device in 2012, an increase of 293 per cent between 2011 and 2012, and a larger increase is expected this year.

“We focused on different features [for mobile] because of the usability perspective,” Mahir said. Tennis Australia determined that people mainly wanted to use the app for scores, schedules and other basic information, he said.

The IBM app is completely separate from the Ticketek app. The mobile ticketing app “caters to purchasing quickly and we don’t want to distract [customers] from that,” he said.

Adam Bender traveled to the Australian Open in Melbourne as a guest of IBM.

Follow Adam Bender on Twitter: @WatchAdam

Follow Computerworld Australia on Twitter: @ComputerworldAU, or take part in the Computerworld conversation on LinkedIn: Computerworld Australia

Tags Australian Openpredictive analyticsTennis AustraliaappsIBMwebsitebig datasocial mediamobiletennisanalyticsscoresCloud

Comments

Comments are now closed

Mixed responses from industry on data retention bill

READ THIS ARTICLE
DO NOT SHOW THIS BOX AGAIN [ x ]