README.md

---
license: mit
task_categories:
- token-classification
task_ids:
- named-entity-recognition
dataset_info:
  features:
  - name: sentence
    dtype: string
  - name: entities
    list:
    - name: end
      dtype: int64
    - name: label
      dtype: string
    - name: start
      dtype: int64
    - name: text
      dtype: string
  - name: data_source
    dtype: string
  splits:
  - name: train
    num_bytes: 859293
    num_examples: 3361
  - name: validation
    num_bytes: 447206
    num_examples: 1819
  download_size: 564719
  dataset_size: 1306499
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
---

---

# Proposed Active Learning Data split
## Overview
This dataset release represents a proposed and experimental data split designed specifically to support and validate planned active learning (AL) cycles for biomedical Named Entity Recognition (NER).

The current version is not a final benchmark split. Instead, it serves as an initial, controlled setup for testing active learning strategies, model uncertainty sampling, 
and iterative annotation workflows prior to large-scale development.

Both splits (train and validation) have been carefully curated to ensure coverage of all three target entity types:

- **CellLine**

- **CellType**

- **Tissue**

This balanced representation is critical for meaningful evaluation of active learning behavior across heterogeneous biomedical entities.


## Dataset Description

Each of the split contain the following features: 
- sentence: list of sentences
- entities: list of dict of entities found in sentences
- data_source: the source of the article where the sentence orginates

The dataset has been curated from three complementary biomedical domains, each contributing distinct entity distributions:
1. Single-Cell transciptomics: rich in CellTypes and Tissues
2. ChembL assay desriptions: rich in CellLines
3. Stem-Cell research: contains all 3 entities

The stem cell–related articles were collected from the CellFinder data repository.
The creation and annotation methodology of the original CellFinder dataset are described in the following reference:

>Mariana Neves, Alexander Damaschun, Andreas Kurtz, Ulf Leser (2012)
>Annotating and evaluating text for stem cell research. 
>In Proceedings Third Workshop on Building and Evaluation Resources for Biomedical Text Mining (BioTxtM 2012),
>Language Resources and Evaluation (LREC) 2012.

## Article PMCIDs and source
### Train Set
| PMCID | In-Split | Source |
|-------|---------|--------|
| PMC12435838 | train | Single-Cell |
| PMC11578878 | train | Single-Cell |
| PMC12396968 | train | Single-Cell |
| PMC11116453 | train | Single-Cell |
| PMC12408821 | train | Single-Cell |
| PMC10968586 | train | CheMBL-V1 |
| PMC10761218 | train | CheMBL-V1 |
| PMC7642379 | train | CheMBL-V1 |
| PMC10674574 | train | CheMBL-V1 |
| PMC1315352 | train | CellFinder |
| PMC2041973 | train | CellFinder |
| PMC2238795 | train | CellFinder |


### Validation Set
| PMCID | In-Split | Source |
|-------|---------|--------|
| PMC12256823 | val | Single-Cell |
| PMC12116388 | val | Single-Cell |
| PMC10287567 | val | Single-Cell |
| PMC12133578 | val | Single-Cell |
| PMC8658661 | val | CheMBL-V1 |
| PMC11350568 | val | CheMBL-V1 |
| PMC12072392 | val | CheMBL-V2/CeLLaTe-V2 |
| PMC12115102 | val | CheMBL-V2/CeLLaTe-V2 |
| PMC2063610 | val | CellFinder |
| PMC1462997 | val | CellFinder |

---

## Intended Use

### Primary Use
- Supervised NER training for biomedical NLP tasks

### Not Intended For
- Clinical or patient-level decision making

---

## Notes and Limitations

- This is an experimental split, subject to change.

- Entity distributions may not reflect real-world prevalence.

- Annotation density varies across domains by design