Publish span mirror

README.md

@@ -0,0 +1,70 @@
+---
+license: cc-by-2.0
+language:
+- en
+tags:
+- tasksource
+- extraction
+configs:
+- config_name: abstracts
+  data_files:
+  - split: train
+    path: abstracts/train.parquet
+  - split: validation
+    path: abstracts/validation.parquet
+  - split: test
+    path: abstracts/test.parquet
+- config_name: papers
+  data_files:
+  - split: train
+    path: papers/train.parquet
+---
+# BioScope negation and speculation spans
+
+Expert annotations of negation and speculation cues and their scopes in biomedical abstracts and full papers (the clinical part, whose text is masked, is left out). Every sentence asks for the negation and speculation cues and the scope of each cue. BioScope has no native split: `abstracts` gets a deterministic document-disjoint 80/10/10 split (`crc32(DocID) % 100`); `papers` has only 9 documents and stays train-only.
+
+Data-only mirror of the [original distribution](https://rgai.inf.u-szeged.hu/sites/rgai.inf.u-szeged.hu/files/bioscope.zip) (SHA256 `b20b525506f7`), documented by
+[bioscope.py](https://huggingface.co/datasets/bigbio/bioscope/blob/43ba162359f01d6436590ba2a014421b6ad09621/bioscope.py). Reproduce with
+[scripts/repackage_dataset/span_structure.py](https://github.com/sileod/tasksource/blob/main/scripts/repackage_dataset/span_structure.py);
+[provenance.json](provenance.json) records hashes, settings and counts. License: cc-by-2.0
+([evidence](https://huggingface.co/datasets/bigbio/bioscope/blob/43ba162359f01d6436590ba2a014421b6ad09621/README.md)). Public biological abstracts and papers only; article texts retain their original terms. Clinical masked text excluded.
+
+Schema: `context`, `question`, `answers` (`text`, `answer_start`: every span, not alternatives) and JSON-string
+`metadata` (`field`, sentence `group`). Used by the `extraction` config of
+[tasksource-jev-typed-decisions](https://huggingface.co/datasets/tasksource/tasksource-jev-typed-decisions).
+
+## Citation
+
+Please cite the original dataset:
+
+```bibtex
+@article{vincze2008bioscope,
+    title = {The {BioScope} corpus: biomedical texts annotated for uncertainty, negation and their scopes},
+    author = {Vincze, Veronika and Szarvas, Gy{\"o}rgy and Farkas, Rich{\'a}rd and M{\'o}ra, Gy{\"o}rgy and Csirik, J{\'a}nos},
+    journal = {BMC Bioinformatics},
+    volume = {9},
+    number = {Suppl 11},
+    pages = {S9},
+    year = {2008},
+    url = {https://doi.org/10.1186/1471-2105-9-S11-S9},
+}
+```
+
+## Citing tasksource
+
+If you use tasksource, its standardized task annotations, or datasets derived from them in published research, please cite the tasksource paper (Sileo, LREC-COLING 2024).
+
+```bibtex
+@inproceedings{sileo-2024-tasksource,
+    title = "tasksource: A Large Collection of {NLP} tasks with a Structured Dataset Preprocessing Framework",
+    author = "Sileo, Damien",
+    booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)",
+    month = may,
+    year = "2024",
+    address = "Torino, Italia",
+    publisher = "ELRA and ICCL",
+    url = "https://aclanthology.org/2024.lrec-main.1361",
+    doi = "10.63317/2h4pkr685mkq",
+    pages = "15655--15684",
+}
+```

abstracts/README.md

@@ -0,0 +1,8 @@
+# abstracts
+
+Public BioScope abstracts, following BigBio.
+Native XML text, nested cue/scope offsets; clinical masked words excluded.
+
+Every sentence is kept; sentences without a cue answer both cue fields with null.
+
+Deterministic document-disjoint split (no native split exists): `crc32(DocID) % 100` assigns each of the 1,273 documents to train (<80), validation (80-89) or test (>=90); every sentence of a document stays in its split.

abstracts/test.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:089d7dd0621f056eefb43abd55573b939b4c9229e110c21e9bb64a5d8cc68589
+size 181822

abstracts/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61053eb8d516d21ce249cb6dc636a74ee87c9571c0677e72c2fb9362ae386ef2
+size 1644103

abstracts/validation.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:606ff1259dac5234f13eec8e4e24444e6ec78ecfb491a7ab3f2047cf77834415
+size 227513

papers/README.md

@@ -0,0 +1,8 @@
+# papers
+
+Public BioScope papers, following BigBio.
+Native XML text, nested cue/scope offsets; clinical masked words excluded.
+
+Every sentence is kept; sentences without a cue answer both cue fields with null.
+
+Train-only: the collection has 9 documents, too few to split representatively.

papers/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a399439f7bfa63a51765d818ff74dea9fa3c38c4af8c96da73f4b6bc557dd049
+size 428933

provenance.json

@@ -0,0 +1,58 @@
+{
+  "source": "bioscope",
+  "mirror": "tasksource/bioscope-spans",
+  "upstream": {
+    "mirror": "tasksource/bioscope-spans",
+    "url": "https://rgai.inf.u-szeged.hu/sites/rgai.inf.u-szeged.hu/files/bioscope.zip",
+    "sha256": "b20b525506f783ab9c120548ac4022ab587bf7cc419beaf759f7253f0cf0b1d1",
+    "revision": "43ba162359f01d6436590ba2a014421b6ad09621",
+    "documentation": "https://huggingface.co/datasets/bigbio/bioscope/blob/43ba162359f01d6436590ba2a014421b6ad09621/bioscope.py",
+    "license": "cc-by-2.0",
+    "license_evidence": "https://huggingface.co/datasets/bigbio/bioscope/blob/43ba162359f01d6436590ba2a014421b6ad09621/README.md",
+    "license_note": "Public biological abstracts and papers only; article texts retain their original terms. Clinical masked text excluded.",
+    "title": "BioScope negation and speculation spans",
+    "about": "Expert annotations of negation and speculation cues and their scopes in biomedical abstracts and full papers (the clinical part, whose text is masked, is left out). Every sentence asks for the negation and speculation cues and the scope of each cue. BioScope has no native split: `abstracts` gets a deterministic document-disjoint 80/10/10 split (`crc32(DocID) % 100`); `papers` has only 9 documents and stays train-only.",
+    "configs": {
+      "abstracts": [
+        "train",
+        "validation",
+        "test"
+      ],
+      "papers": [
+        "train"
+      ]
+    },
+    "citation": "@article{vincze2008bioscope,\n    title = {The {BioScope} corpus: biomedical texts annotated for uncertainty, negation and their scopes},\n    author = {Vincze, Veronika and Szarvas, Gy{\\\"o}rgy and Farkas, Rich{\\'a}rd and M{\\'o}ra, Gy{\\\"o}rgy and Csirik, J{\\'a}nos},\n    journal = {BMC Bioinformatics},\n    volume = {9},\n    number = {Suppl 11},\n    pages = {S9},\n    year = {2008},\n    url = {https://doi.org/10.1186/1471-2105-9-S11-S9},\n}"
+  },
+  "settings": {
+    "qasrl_partition": "orig",
+    "qasrl_validation": ">=2 valid human judgments and strict majority; modal complete reading",
+    "bioscope_members": [
+      "abstracts.xml",
+      "full_papers.xml"
+    ],
+    "bioscope_split": "abstracts: deterministic document-disjoint crc32(DocID)%100 80/10/10; papers: train-only (9 documents)",
+    "uploads": false
+  },
+  "rows": {
+    "abstracts/train": 22563,
+    "abstracts/validation": 3045,
+    "abstracts/test": 2512,
+    "papers/train": 6384
+  },
+  "counts": {
+    "abstracts/train": {
+      "input_sentences": 9535
+    },
+    "abstracts/validation": {
+      "input_sentences": 1266
+    },
+    "abstracts/test": {
+      "input_sentences": 1070
+    },
+    "papers/train": {
+      "input_sentences": 2670
+    }
+  },
+  "script": "scripts/repackage_dataset/span_structure.py"
+}