pnr-svc/Drugs-NER-Data

Named Entity Recognition (NER) Dataset

Overview

This dataset is designed for Named Entity Recognition (NER) tasks. It contains annotated text data with entities classified into various categories. The dataset is split into training, validation, and test sets, and is suitable for training and evaluating NER models.

Dataset Description

Features

The dataset includes the following features:

• —tokens: A sequence of strings representing the words or tokens in the text.
• —ner_tags: A sequence of integers representing the NER tags assigned to each token. The tags correspond to specific named entities in the text.

Label Mapping

The ner_tags feature uses a set of predefined labels mapped to integers as follows:

• —0: B-AIDS
• —1: B-AVBlock
• —2: B-AbnormalBloodClotting
• —3: B-AbnormalBloodTests
• —4: B-AbnormalKidneyFunction
• —5: B-AbnormalLiverFunction
• —6: B-AbnormalLiverFunctionTests
• —7: B-AbnormalVaginalBleeding
• —8: B-Accidents
• —9: B-Aches
• —... (and so on) ...
• —1610: O

Each integer represents a specific label used to annotate the entities in the text.

Splits

The dataset is divided into three splits:

• —Training Set: 80% of the dataset, used for training the NER models.
• —Number of Examples: 14,205
• —Size: 8,592,704 bytes

• —Validation Set: 10% of the dataset, used for tuning hyperparameters and validating the model.
• —Number of Examples: 5285

• —Test Set: 10% of the dataset, used for evaluating the final performance of the model.
• —Number of Examples: 5285

Usage

To load the dataset, use the datasets library as follows:

from datasets import load_dataset

dataset = load_dataset("pnr-svc/Drugs-NER-Data")