Johansson, RichardJohansson, Richard, editor2025-05-142025-05-142025https://hdl.handle.net/10062/109659I resurrected a dataset containing annotated event-to-event temporal relations in Swedish traffic accident reports, and then benchmarked how well contemporary large language models classify these relations. Can the current generation of LLMs outperform our results from two decades ago?enAttribution-NonCommercial-NoDerivatives 4.0 Internationalhttps://creativecommons.org/licenses/by/4.0/Book