Covering Disruptive Technology Powering Business in The Digital Age

WIMBLEDON fans are being served a mash up of machine learning and advanced analytics in a bid to capture viewers’ attention on social media and digital platforms.

Statistics and analytics have been a big feature of grand slam tennis for some years now. But what’s new this year is Watson.

IBM’s flagship AI-driven analytics platform has been tasked with crunching through the hundreds of thousands of social media and online posts which the event generates.

It’s mission will be to find the stories that fans are most engaged with, and drive creation of the sort of content that they most want to see.

And crucially, for the first time this year thanks to Watson’s machine learning capabilities, this will be a predictive process, rather than an attempt to piggyback trends once they occur.

Alexandra Willis, head of communications, content and digital at the All England Lawn Tennis and Croquet Club which hosts the yearly grand slam, agreed to talk to me about some of the changes that have been put in place this year.

Of the integration between Watson and social media, She told me “This allows us to not just look at and respond to trends, but to actually pre-empt them. We’re hoping this will help in our quest, not necessarily to always be first but certainly to be early into the conversation when critical things are happening.”

As an example of what Watson could be able to do, Willis recalls 2014 when three Canadian players – Milos Raonic, Eugine Bouchard and Vasek Pospisil – all reached the semi-finals of major tournaments.

This generated, unexpectedly, a lot of conversation about Canadian tennis, which broadcasters and media were forced to engage with reactively. “A lot of people were asking ‘where has this come from?’ ‘Is it due to something specific?’ so we were able to adapt our content to make sure we were answering these questions,” Willis says.

In theory, working predictively, Watson will be able to spot emerging trends – such as an unexpectedly good performance by players from a particular nation – before they start to trend on Twitter.

“We will hopefully be able to monitor the particular interest in a particular court, or if there is one player garnering particular interest we will be able to hop on and pre-empt that trend.”

Although Wimbledon will be using predictive analytics and machine learning to suggest what content will be most appreciated by fans, it isn’t yet planning to go down the route of automating the creation of that content.

Social media posts, alerts and reports will still be hand crafted by the content team, and of course the balance between content and context has to be judged correctly. “No we don’t do that”, Willis tells me, “Yes, it’s important for us to provide interesting insight from data but it’s also really important that it’s done in the correct tone for Wimbledon and with the correct approach.

“There is still that editorial judgement at the heart of it, and it’s still very much an authentic experience.

“We don’t want to turn into a newswire and we don’t want to just have a stream of statistical information spiralling out there.”

Fans this year will also see an enhanced integration of the IBM Slamtracker statistics interface into the tournament’s media output.

Rather than running as a standalone app within the website, as has previously been the case, insights from the system will be used across all channels, shareable across social media and embedded into match reports. “We’re not having it as this standalone area for people who are interested in statistics, it’s actually something that becomes much more relevant and meaningful for everybody.”

Interestingly despite the array of high tech sensor, motion capture and recording technology deployed, IBM – which as the Official Technology Supplier to The Championships handles Wimbledon’s data operations in-house – still stations two human data collectors on each court for each match.

They provide a human standard of commentary and reporting that can pick up some nuances that are still beyond today’s automated sensors – for example the difference between a player making a forced error or an unforced error.

These people are generally tennis experts who have been specially trained as data operatives – which is reportedly an easier process than training data experts to become tennis experts.

This article was originally published on www.forbes.com and can be viewed in full

(0)(0)