This research introduces NarrativeReason, a novel dataset that enhances LLMs' temporal reasoning for better summarization of event sequences in social media threads.

• Focuses on understanding sequential relationships between multiple events rather than just pair-wise connections
• Demonstrates improved timeline summarization capabilities in LLMs when trained with temporal reasoning tasks
• Particularly effective for extracting chronological narratives from unstructured social media content

In medical applications, this advancement enables better monitoring of mental health trends through social media analysis, potentially allowing for earlier intervention by identifying developing patterns in patient narratives over time.

Temporal reasoning for timeline summarisation in social media