Data Story Finder – Artificial Intelligence supports live commentary

Photo: DFL/Getty Images/Simon Hofmann

13 October 2021 – The people who deliver the live commentary during football matches must be good at multitasking: not only do they have to follow closely what is happening on the pitch; they are also expected to describe the action skilfully and provide relevant background information so viewers can grasp the wider context. To enable commentators to instantly recognise surprises, unusual accomplishments or even new records, a new Artificial Intelligence-driven assistance system is now available.

Keeping an eye on key information: the Commentary Live System

The “Commentary Live System“ (CLS) is an editorial support system available to media partners of DFL for Bundesliga and Bundesliga 2 matches. It delivers a wide range of supplementary data, from match facts and stats to lineups, and through to individual player details such as the proper pronunciation of names, various rankings, and other background knowledge. Commentators can incorporate all this information into their commentary as they see fit. What is more, the system allows them to enter notes of their own ahead of a match for subsequent use.

Additional statistical information is displayed in a live ticker format, assembled by the data editors working in the back office to select and edit relevant match, seasonal and historical statistics. The presentation of all these data helps the commentators provide additional context for the match in progress, such as whether or how often specific situations have occurred in the past, or whether a player’s performance is particularly noteworthy.

Artificial Intelligence in action: contextual data delivered in real time

Bundesliga Match Facts powered by AWS have elevated the presentation of context-related live information to an entirely new level. Using Artificial Intelligence, this feature captures, calculates and visualises contextual data in real time to help viewers better understand what is happening on the pitch. A new, truly innovative service has been added for the 2021/2022 season: the Data Story Finder, developed jointly by the DFL subsidiary Sportec Solutions and Amazon Web Services (AWS). As a supplementary source of information within the CLS, it accelerates the delivery of context significantly.

Using smart algorithms, the Data Story Finder searches for matching “stories”, i.e., it correlates the live match data captured automatically in real time with other match, seasonal or historical data, then presents the results to the data editors as an additional source of information. The editors evaluate these “stories” and decide whether to feed them into the live ticker, categorised as either match and competition-related or player and team-related insights. For example, stories in the Average Positions category analyse whether a team’s current positioning strategy differs significantly from the season to date.

Millions of data points evaluated in an instant

The event and position data (the latter alone comprising 3.6 m data points per match) are provided by Sportec Solutions and converted into Bundesliga Match Facts and other formats. The algorithms process this information using a set of currently 15 pre-defined story templates, each of which evaluates specific parameters and enters the results into a pre-worded text statement. These text elements are available in German and English so international correspondents can use the same system.

For example, there is a “Shot Efficiency” story which evaluates scoring chances to determine whether a win is more or less surprising (the text in the ticker might read: “While Team A scored 0.74 fewer goals than expected, Team B scored 0.88 higher than expected.”) A “Most Pressed Player” story shows which player is experiencing especially high pressure from the opposing team while in ball possession, and how this relates to the average pressure his team has been exposed to. The ticker might show a message like “Marco Reus experienced pressure 25 times, 110per cent more frequently than the average of Borussia Dortmund”. Other stories relate to the passing profile, attacking zones or expected goals. Plans are to further expand the portfolio to deliver even more context information on player and team performance.