QMSum
A query-focused meeting summarization benchmark dataset released by Zhong et al. in 2021. It contains approximately 230 meeting transcripts spanning academic meetings, product meetings, and parliamentary debates, each annotated with human-labeled query–summary pairs. It is a widely used benchmark in the query-focused summarization field.
Timeline (last 90 days)
QMSum is the mainstream benchmark dataset for query-focused meeting summarization tasks, but it does not provide a standardized scorer.
The QMSum dataset was released by Zhong et al. in 2021 and contains approximately 230 meeting transcripts drawn from three settings—academic meetings, product meetings, and parliamentary debates—each annotated with human-labeled query-summary pairs.
The researchers re-scored or re-generated results for 15 systems under a unified implementation framework.
All Facts (3)
QMSum is the mainstream benchmark dataset for query-focused meeting summarization tasks, but it does not provide a standardized scorer.
50%UnverifiedThe QMSum dataset was released by Zhong et al. in 2021 and contains approximately 230 meeting transcripts drawn from three settings—academic meetings, product meetings, and parliamentary debates—each annotated with human-labeled query-summary pairs.
50%UnverifiedThe researchers re-scored or re-generated results for 15 systems under a unified implementation framework.
50%