barthfab commited on
Commit
04b6193
·
verified ·
1 Parent(s): 264a5e9

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +64 -0
README.md CHANGED
@@ -36,3 +36,67 @@ configs:
36
  - split: not_parallel
37
  path: data/not_parallel-*
38
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
36
  - split: not_parallel
37
  path: data/not_parallel-*
38
  ---
39
+
40
+ # Dataset Card: PISA Multimodal (Parallel & Not-Parallel)
41
+
42
+ ## Summary
43
+ This dataset contains **48 parallel multimodal samples** (paired TXT↔PDF) derived from **PISA studies up to 2012**, plus **47 non-parallel samples** (TXT-only or PDF-only). Each sample may include multiple questions. Content is available in **German** and **English**.
44
+
45
+ **Source & usage:** Materials are published by the **OECD** and are provided here **for non-commercial use** only. Please verify that your usage complies with OECD terms.
46
+
47
+ **Quality note:** Some of the **parallel** samples are **noisy** (e.g., OCR artifacts or imperfect alignments). These are **explicitly labeled in the filename** so they can be filtered out during experiments.
48
+
49
+ ## Organization & Splits
50
+ - **`train`** — *parallel* split: 48 samples (TXT and matching PDF).
51
+ - **`not_parallel`** — 47 samples where only TXT **or** only PDF is present.
52
+ *Note:* In this split, one of `Content` or `PDF` can be `null`.
53
+ *(Counts reflect the provided corpus; they may change if you add/remove files.)*
54
+ ## Languages
55
+ - **German:** `deu_Latn`
56
+ - **English:** `eng_Latn`
57
+
58
+ ## Tasks
59
+ - **Reading** comprehension
60
+ - **Math** problem solving
61
+
62
+ ## Data Fields (Columns)
63
+ - **`ID`** *(int32)* — Running index of the sample (per build script).
64
+ - **`Task Name`** *(string)* — Base name of the material (filename stem before the last underscore), e.g., `Bookshelves`.
65
+ - **`PDF`** *(Pdf feature)* — Full PDF object (stored via `datasets.Pdf`). May be `null` in `not_parallel` if missing.
66
+ - **`Language`** *(string)* — Language code of the folder, e.g., `deu_Latn` or `eng_Latn`.
67
+ - **`Task`** *(string)* — Task category from the folder name: `reading` or `math`.
68
+ - **`Content`** *(string)* — Full text from the paired `.txt`. May be `null` in `not_parallel` if missing.
69
+ - **`TXT Filename`** *(string)* — Basename of the associated `.txt` file (if present).
70
+ - **`PDF Filename`** *(string)* — Basename of the associated `.pdf` file (if present).
71
+
72
+ ## Source Data & Provenance
73
+ - **Origin:** Items compiled from **OECD PISA** materials (≤2012).
74
+ - **Structure:** Files are organized by `<language>_<task>` folders, e.g., `german_reading`, `english_math`.
75
+ - **Derivation:** Parallel pairs were formed by matching TXT and PDF stems; the `not_parallel` split contains items missing one counterpart.
76
+
77
+ ## Licensing & Usage
78
+ - **License:** *Other / Non-commercial.* The content is published by the **OECD**. Redistribution and use should comply with **OECD non-commercial terms**. This card is not legal advice—please review the official terms for your use case.
79
+
80
+ ## How to Load
81
+ ```python
82
+ from datasets import load_dataset
83
+ # From the Hub (replace with your repo id)
84
+ ds = load_dataset("your-username/your-dataset")
85
+ # Parallel split
86
+ train = ds["train"]
87
+ # Not-parallel split
88
+ not_parallel = ds["not_parallel"]
89
+ # Example access
90
+ row = train[0]
91
+ print(row["Task Name"], row["Language"], row["Task"])
92
+ # PDF is a Pdf feature record with a path/bytes:
93
+ print(row["PDF"])
94
+ ```
95
+
96
+ ## Intended Uses & Limitations
97
+ - **Intended:** research on multimodal QA, OCR+NLP, cross-lingual evaluation, dataset tooling.
98
+ - **Not intended:** commercial exploitation of OECD content; deployment without verifying rights.
99
+ - **Limitations:** Some samples include multiple questions per PDF/TXT; `not_parallel` items are incomplete by design; some parallel items are flagged as **noisy** in filenames.
100
+
101
+ ## Citation
102
+ If you use this dataset, please cite the OECD PISA materials appropriately and reference this dataset card.