Today, media companies try to increase their advertising value by directly addressing target groups and tailor-made content in social networks. Artificial Intelligence (AI) offers completely new approaches here, as the cognitive assistant Watson from IBM shows. The AI system combines results from video, audio and data analysis to generate the most exciting and important moments of the Wimbledon Championships.
By Steffen Konrath
Wimbledon is the Mecca of tennis fans and their athletes. Whoever plays here, has it. Whoever watches here is as tense as the string of a racket. And that’s what A.I. is taking advantage of. Facial expressions, cheers and curses reflect the emotions of the players and spectators; if these human reactions are recorded in the stadium, one can assume that the spectator at home would also like to be shown this moment. Wimbledon’s collaboration with IBM Watson focuses on analyzing these highly emotional moments on site. In short videos, the AI combines moments full of excitement, disappointment and joy into highlight packages. This not only minimizes the evaluation of hours of video material for television broadcasts – it also feeds the social networks.
Wimbledon has the world’s largest social media fan community
The video highlights generated by Watson have been and will continue to be used in 2018 to stimulate discussion and content on social channels such as Twitter, Facebook and Telegraph.co.uk. Through this type of social media monitoring, Wimbledon’s editorial team not only succeeds in analyzing the reaction of viewers, but also actively challenges them – and thus, by increasing their involvement, significantly increases advertising revenues. With a fan base of over 4.1 million people, Wimbledon has more Facebook, Twitter and Instagram followers than The Australian, the French Open and US Open. The French Open ranks second with a comparable social media fan base of 3.7 million. According to Twitter Buzz, the 2014 Wimbledon Final reached a climax in social media campaigns with over 585,000 posts compared to similar sporting events. This number was almost twice as high as at any other Grand Slam tournament. In addition, with each response to the AI-based outsourcing of content, information flows back into the system and refines the underlying model, so that it continues to improve over time.
Combination of player emotions, viewer reactions and score
„All England Lawn Tennis & Croquet Club Limited“ AELTC has been working with AI in the evaluation and transmission of game data since 2015 – a development that was and still is being driven forward in the following years. Last year, the team began using artificial intelligence to create videos with great success: This year, over 14 million of these videos were viewed on the social network. Further milestones are planned for 2018. With the latest methods, “player emotions” can now also be recognized and processed in video recordings. The AI combines various factors such as playing skills, personality traits and behaviors – for example cheering or anger – of the respective player. In addition, Watson is trained to create the videos by analyzing crowd noise, player movements and match data. Using this data, Watson selects the most important scenes from a game.
Neural network “SoundNet” helps Watson learn
IBM recently expanded its solution to include SoundNet, a deep neural network that extracts relevant features from video data. Metadata such as sounds, movements and the like are linked to the respective score. In addition, real-time data from other sources are integrated, including statistics, referee decisions, ball position and player location. The opinions of fans from social networks such as Twitter are also included. Because IBM’s cognitive assistant can extract and interpret the most important content from these components, it is able to summarize the highlights of the game in a short video.
2018 will bring an AI Highlights Dashboard
In this Wimbledon season, an AI Highlights dashboard was added that was able to process any worth seeing shot of a game in no time at all. The cross-platform application also offers the current score and statistics for all running games as well as corresponding insights. Fans can also see the tactics of the athletes before the game.
This data can also be used for match forecasts: To make predictions, Watson has been “fed” with data from players and game tactics from 22 years of Wimbledon history since 2015 – with this background knowledge, AI (Artificial Intelligence) in conjunction with the current game data and behavior patterns can constantly evolve.
AI Pavilion with Expo & Conference @MEDIENTAGE 2018
The topic AI is a focal point of MEDIENTAGE 2018.
In the MEDIENTAGE AI Pavilion Expo area, digital and tech companies, media houses, service providers and start-ups will present their innovations, technologies and services relevant to the media business of the future. In addition, research institutions will show future-oriented prototypes that deal with the creative potential of AI.
The AI Expo (24-26 October) will be accompanied by the two-day congress program of the AI Conference (24 and 25 October), at which leading AI experts will offer an outlook on upcoming trends, technologies and usage scenarios, present use cases from the media world and provide concrete insights into their strategies and learnings in the implementation of AI innovations.
All information about the AI Pavilion can be found here.
You want your company to be present in the AI Pavilion? The key facts about the MEDIENTAGE 2018 can be found here. Please contact us for further information – we look forward to hearing from you. Mail to: Hans.Haeusler@medientage.de