Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | https://arxiv.org/abs/2501.03403 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911313361371136 |
|---|---|
| author | Giovannini, Simone Coppini, Fabio Gemelli, Andrea Marinai, Simone |
| author_facet | Giovannini, Simone Coppini, Fabio Gemelli, Andrea Marinai, Simone |
| contents | We present a unified dataset for document Question-Answering (QA), which is obtained combining several public datasets related to Document AI and visually rich document understanding (VRDU). Our main contribution is twofold: on the one hand we reformulate existing Document AI tasks, such as Information Extraction (IE), into a Question-Answering task, making it a suitable resource for training and evaluating Large Language Models; on the other hand, we release the OCR of all the documents and include the exact position of the answer to be found in the document image as a bounding box. Using this dataset, we explore the impact of different prompting techniques (that might include bounding box information) on the performance of open-weight models, identifying the most effective approaches for document comprehension. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2501_03403 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | BoundingDocs: a Unified Dataset for Document Question Answering with Spatial Annotations Giovannini, Simone Coppini, Fabio Gemelli, Andrea Marinai, Simone Computation and Language Artificial Intelligence We present a unified dataset for document Question-Answering (QA), which is obtained combining several public datasets related to Document AI and visually rich document understanding (VRDU). Our main contribution is twofold: on the one hand we reformulate existing Document AI tasks, such as Information Extraction (IE), into a Question-Answering task, making it a suitable resource for training and evaluating Large Language Models; on the other hand, we release the OCR of all the documents and include the exact position of the answer to be found in the document image as a bounding box. Using this dataset, we explore the impact of different prompting techniques (that might include bounding box information) on the performance of open-weight models, identifying the most effective approaches for document comprehension. |
| title | BoundingDocs: a Unified Dataset for Document Question Answering with Spatial Annotations |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2501.03403 |