by Noah D'Mello

Analyzing your social media sentiments, one ball at a time

Oct 20, 2015
AnalyticsEnergy IndustryGovernment

From analyzing sports tweets to tracking crowd movements in large venues, IBM Research-India hit a six when it integrated social media and analytics to make sports more engaging and interesting.u00a0

Wimbledon 2015 Men’s Finals, Roger Federer vs Novak Djokovic: The second set tie-breaker became the most tweeted game in the entire tournament—Why shouldn’t it? It is touted as one of the best Wimbledon tiebreakers. You may have noticed that everyone around you took to Facebook and Twitter to show either their disappointment or their happiness. During such games, going through all your social media feeds is entertaining. But leveraging this data to create more buzz is something that technology has the ability to do. Please welcome to the courts: Social media analytics.

In February 2015, 1.3 billion Indians had their eyes on Team India, who were the defending champions at the 2015 ICC World Cup. IBM Research in Bangalore took advantage of the fact that social media citizens, both active and inactive, will now use networking sites more than ever to express themselves. In anticipation to this, IBM entered the sports pavilion with Score With Data, a social newsroom on Twitter that analyzed every tweet and hashtag related to the World Cup, and thus scrutinized your social sentiments and emotions during India’s run at keeping their hands on the cup.

“Social media is such an integral part in the way people enjoy every experience,” Sriram Raghavan, Senior Manager, Information & Analytics Departments, IBM Research-India, said. “The idea behind Score With Data was to use large-scale analytics capabilities that can understand data to make fan experience in a major sporting event like the World Cup more interactive, engaging, fun, and insightful.”

Imagine this: Hordes of crazy Indian fans are tweeting about the fourth ball in the 40th over, which has now become the turning point of the match—India’s victory now seems certain. IBM was able to detect these turning points, which were micro events, even though all the “devotees” did not use the same words or language to “pray” to their “gods” on Twitter.

The team behind Score With Data, a showcase of IBM’s analytic capabilities, analyzed such sentiments on social media—about 3 million tweets in 10-minute intervals and 15 million social media posts every day. This allowed them to summarize what Indians felt, who were their favourite players, who were they disappointed in the most, among various other feelings. Twitter cards, which allow summarization of data, were used, which provided 6 percent engagement, Raghavan said.

“The focus was mostly on the fact that folks in research and analytics capabilities could leverage IBM’s existing platforms and customize it for the fans,” he added.

The team introduced the Social Sentiment Index, which used Watson technologies to make sense of text. “Our Watson capability could make sense of somebody saying ‘Kitna accha chhaka tha! (That was such a great sixer!)’,” Raghavan said.

To come up with results for this index, Watson pulled out sentiments from Twitter about events, players, tournaments, facilities—everything to do with the World Cup—and presented it using interesting dashboards.

IBM also partnered with Wisden India, which had tons of historical data, to add context to the sentiments during the matches. Raghavan said that the Wisden Impact Index, which was an application on IBM’s Bluemix environment, was “driven by the fact that if we combine lots of historical data with data from the game, then it will make some interesting analysis.”

This index allowed IBM to summarize player performance such as who is going to be impactful and who has been the most important player in past games.

All this analytics translated to a Twitter reach of 9.1 million, 2.1 million profile visits, and 29,100 followers on Score With Data’s Twitter handle in 45 days.

Diluting this information and making it available to the people in the simplest form is never an easy task. For this, IBM partnered with CNN-IBN, which broadcast this information on their show “Kings of Cricket.”

The show, which had around 30 episodes and a viewership of around 300,000 viewers, complemented the data provided by IBM and also functioned like a match summary.

Social media analytics hits a backhand winner

It is no surprise that IBM has been delivering technology solutions to tennis grand slam tournaments for decades now. But because of the social media explosion around the world, priorities are slowly changing.

Raghavan said, “The shift is now more and more toward engagement, both physically at the event and at the websites.”

During a match you may ask: What is Andy Murray’s second serve percentage against Novak Djokovic this year compared to the previous year where he has played well against him? “You can pull out this information in a jiffy and it becomes very powerful for the fan to see,” he said.

This was realized at Australian Open 2015, where IBM redesigned the Slam Tracker to make it more robust and with an enhanced visualization. This year the Slam Tracker provided detailed ball-by-ball statistics and player movements to summarize data in a more consumable way for experts and commentators at the Australian Open.

“This was a combination of analysis of real-time data and grand slam data of over 8 years,” he said.

Melbourne Park, home to the Australian Open, has around 22 outdoor courts and planning your day accordingly is important because you do not want enthusiastic fans missing out an unexpected match which may not have seemed interesting at the start but has turned out to be crowd puller.

Crowd Tracker, developed by IBM, exploited the fact that the venue had a number of pre-installed WiFi hotspots, which can help track crowd movement. This gives organizers a chance to tell fans who are not present at any match to hop over to a court which could be a one of the most thrilling matches.

Australian Open has around 600,000 people who attend live matches. But Tennis Australia’s website has over 15 million visitors.

“The important thing is to handle such a huge number at tremendous peak loads… So being able to manage this as an IT infrastructure is the key,” he said.

“You can go the traditional way and say that finals and semifinals will have more viewership… But with the ability to apply Watson-like technology, this year we anticipated that just with the chatter on social media or our website about specific matches that can give a load, we can provision enough resources so that stats about that match will go in real time to all the people who are watching the match,” he added.

IBM also hit winners in this year’s Wimbledon tournament, the oldest and widely considered the most prestigious grand slam tournament. The analytics team pulled out information from various social media sites and identified the most engaging and influential person online. These influential people not included sports personalities but also celebrities such as writers and actors.

“The team built an automatic algorithm that can process data in real time, millions of tweets per day. It combines predictive analytics and graph analytics that became the part of Wimbledon enjoyed by fans,” Raghavan said.

Such data could be used to estimate the online reach of the matches and the content that is most important to the fans, which may help content developers to bring out interesting analysis of the matches played at the All England Club.

A match is not over even after the last shot is played. And in this current day and age of extensive social media engagement, analytics is doing its best to keep the memories of the game alive even after the players have walked out of the pavilion and are basking in their victory, knowing that their fans are still talking about them.