Researchers introduce QFES task and QFESum dataset for query-focused event summarization

Researchers propose the Query-Focused Event Summarization (QFES) task to address the lack of event-oriented summarization in existing datasets. They construct the QFESum dataset, which contains 8 thematic events, 16,684 documents, and 104 queries.

• The study introduces a two-stage framework comprising Query-Focused Retrieval with Adaptive Thresholding (RAT) and Query-Focused Summarization based on Hierarchical Clustering (SHC).
• Experimental results demonstrate that RAT and SHC consistently outperform baselines on the QFESum benchmark.
• The dataset and code are publicly.