This dataset is part of the KoViDoRe Benchmark for Korean visual document retrieval evaluation. Specifically, it is focused on Office Documents such as reports and press releases for text-to-image retrieval tasks.

Dataset Summary

Domain: Korean business/office documents (e.g., 보고서, 보도자료 등) Task: Visual document retrieval (text query → relevant document image) Format: BEIR-compatible (corpus, queries, qrels)

Statistics:

Number of Documents… See the full description on the dataset page: https://huggingface.co/datasets/whybe-choi/kovidore-office-v1.0-beir.