Querying Football Matches for Event Data: Towards Using Large Language Models

Football, being one of the most popular sports in the world, has attracted significant attention from researchers exploring the potential of Artificial Intelligence (AI). In particular, Large Language Models (LLMs), exemplified by digital assistants such as ChatGPT, have proven their capabilities and offer a potentially effective avenue for football research. However, accessibility of football data remains a challenge, as the datasets collected by providers are often inaccessible. This case study presents a proof-of-concept that addresses this challenge by introducing an innovative web scraping approach to extract football event data and making it accessible e.g. for scientific research with LLMs. To this end, the extracted data is structured into coherent sentences for linguistic compatibility. The results show the successful integration of LLMs with football event data, enabling the extraction of information through retrieval-augmented generation. This work makes a first contribution to the field by bridging the gap between football and LLMs, demonstrating the potential for further analysis.