Dataset Viewer
Auto-converted to Parquet Duplicate
dataset_id
stringlengths
3
22
pdf_id_x
int64
0
7
pdf_name
stringclasses
8 values
pdf_title
stringclasses
8 values
page_number
int64
1
679
base64_str
stringlengths
24.5k
11.9M
base64_bytes
int64
24.5k
11.9M
transcription
stringlengths
45
7.45k
transcription_input_tokens
float64
0
850
transcription_output_tokens
float64
0
1.62k
pdf_id_y
null
error
null
1_1
1
1.pdf
Journal of Geophysics
1
"iVBORw0KGgoAAAANSUhEUgAACbAAAA21CAIAAABqgr3/AAAACXBIWXMAAC4jAAAuIwF4pT92AAVvzklEQVR4nOzdB3wVVd7/8Vj(...TRUNCATED)
475,176
"Cc J 3} NIEDERSACHSISCHE STAATS- UND ~ L UNIVERSITATSBIBLIOTHEK GOTTINGEN Werk Jahr: 1980 Kollek(...TRUNCATED)
850
277
null
null
1_2
1
1.pdf
Journal of Geophysics
2
"iVBORw0KGgoAAAANSUhEUgAACgYAAA3CCAIAAABmrx35AAAACXBIWXMAAC4jAAAuIwF4pT92AAdlOUlEQVR4nOzde0CUVf74cbB(...TRUNCATED)
646,324
"Journal Gy Geophysics JA SUSC AIA TLR Geophysik Volume 48 1980 Managing Editors W. Dieminger J.(...TRUNCATED)
850
170
null
null
1_3
1
1.pdf
Journal of Geophysics
3
"iVBORw0KGgoAAAANSUhEUgAACMUAAAw6CAIAAADy/YFaAAAACXBIWXMAAC4jAAAuIwF4pT92AA4Vg0lEQVR4nOzdd7xV1Z0/7ms(...TRUNCATED)
1,230,788
"Journal of Geophysics — Zeitschrift fur Geophysik This journal was founded by the Deutsche Geoph(...TRUNCATED)
850
608
null
null
1_4
1
1.pdf
Journal of Geophysics
4
"iVBORw0KGgoAAAANSUhEUgAACgYAAA3CCAIAAABmrx35AAAACXBIWXMAAC4jAAAuIwF4pT92ABTMYElEQVR4nOzdeXwV1f3/8QQ(...TRUNCATED)
1,817,492
"Author Index Alekseev A.S. 161 173 Baranskiy L.N. 1 Baumgardt D.R. 124 Baumjohann W. 7 Bock (...TRUNCATED)
850
937
null
null
1_5
1
1.pdf
Journal of Geophysics
5
"iVBORw0KGgoAAAANSUhEUgAACgYAAA3CCAIAAABmrx35AAAACXBIWXMAAC4jAAAuIwF4pT92AA2xk0lEQVR4nOzdebxVVd0HfsB(...TRUNCATED)
1,196,676
"IV Inverse Seismic Problems A Characteristic Method for Numerical Solution of the Inverse Kinemat(...TRUNCATED)
850
558
null
null
1_6
1
1.pdf
Journal of Geophysics
6
"iVBORw0KGgoAAAANSUhEUgAACMQAAAxmCAIAAABmcyLoAAAACXBIWXMAAC4jAAAuIwF4pT92AIfffUlEQVR4nNy9aZPjSLIkWP/(...TRUNCATED)
11,872,868
"Journal of | Geophysics Zettschrift fur Geophysik Volume 48 Number1 1980 ledersGchsische Staats- (...TRUNCATED)
850
55
null
null
1_7
1
1.pdf
Journal of Geophysics
7
"iVBORw0KGgoAAAANSUhEUgAACOAAAAxECAIAAABqoSV/AAAACXBIWXMAAC4jAAAuIwF4pT92AAs2OklEQVR4nOy9vY4VSfK4fTQ(...TRUNCATED)
979,808
"Journal of Geophysics — Zeitschrift fur Geophysik Edited for the Deutsche Geophysikalische Gesel(...TRUNCATED)
850
1,084
null
null
1_8
1
1.pdf
Journal of Geophysics
8
"iVBORw0KGgoAAAANSUhEUgAACMQAAAw6CAIAAAAdP+pkAAAACXBIWXMAAC4jAAAuIwF4pT92AA0HtElEQVR4nOzdfYxd5X0n8PG(...TRUNCATED)
1,138,692
"J. Geophys. 48 1-6 1980 Original Investigations Journal of Geophysics The Analysis of Simultane(...TRUNCATED)
850
1,071
null
null
1_9
1
1.pdf
Journal of Geophysics
9
"iVBORw0KGgoAAAANSUhEUgAACMQAAAw5CAIAAACbq5jKAAAACXBIWXMAAC4jAAAuIwF4pT92ADPqx0lEQVR4nOy9ebzV4/r/f84(...TRUNCATED)
4,536,692
"Table 1. List of stations Station Nomen- Geo- Geo- L-para- Local clature magnetic magnetic meter (...TRUNCATED)
850
925
null
null
1_10
1
1.pdf
Journal of Geophysics
10
"iVBORw0KGgoAAAANSUhEUgAACMQAAAw5CAIAAACbq5jKAAAACXBIWXMAAC4jAAAuIwF4pT92AC4GBUlEQVR4nOzdZ3RV1dr/fdJ(...TRUNCATED)
4,021,700
"18 Oct 1974 A Midnight 20F AQT LIN KNG T/A ry) HEL 1 G ; te ns 5 é : - pe \\\\e] {t u is \\\\g (...TRUNCATED)
850
733
null
null
End of preview. Expand in Data Studio

GDZ Scientific Document Retrieval Benchmark

A needle‑in‑a‑haystack benchmark for scientific document retrieval, built from historical volumes of the Göttinger Digitalisierungszentrum (GDZ). This dataset explicitly adapts the IRPAPERS methodology onto a real‑world, multilingual corpus to evaluate both text-based and visual document retrieval models.

Dataset Structure

The dataset is divided into two operational configurations:

1. queries

Contains the curated evaluation question set.

  • Fields: pdf_id, page_number, question, answer
  • Size: 180 needle-in-the-haystack questions.

2. docs

Contains the complete target corpus text and image database.

  • Fields: dataset_id, pdf_id_x, pdf_name, pdf_title, page_number, base64_str, base64_bytes, transcription, transcription_input_tokens, transcription_output_tokens, pdf_id_y, error
  • Size: ~3,000 document pages.
  • Format: Tab-separated value format compressed via Gzip (.csv.gz), ensuring raw text alignment and completely unquoted ("") Base64 image strings.

How It Was Built

  1. Corpus Selection: Selected historical geophysics & mathematics volumes from the GDZ containing complex scientific text, tables, formulas, and diagrams (English + German).
  2. Visual Processing: Extracted high-resolution page images (300 DPI) and encoded them into raw Base64 layout feature matrices (base64_str).
  3. Text Extraction: Processed all pages through Tesseract OCR to build a parallel searchable text layer (transcription).
  4. Query Generation: Generated and manually filtered 180 highly specific verification questions targeting fine-grained methodological parameters and equations buried across the ~3,000 pages.

Usage

You can load either configuration directly into your evaluation pipeline using the Hugging Face datasets library:

from datasets import load_dataset

# 1. Load the 180 question evaluation set
queries_dataset = load_dataset("Trungdaik/Visual_information_retrieval", "queries", split="train")
print("Sample Query:", queries_dataset[0])

# 2. Load the ~3,000 target document pages corpus
docs_corpus = load_dataset("Trungdaik/Visual_information_retrieval", "docs", split="train")
print("Sample Target Page:", docs_corpus[0]["dataset_id"])
Downloads last month
48