farahbs/digit_extraction

Dataset Card for Dataset Name

Dataset Details

Dataset Description

This dataset is designed to extract and record individual digits from numbers. For each number, it contains descriptions of the digit being extracted, such as the 1st digit, 2nd digit, etc., and the corresponding extracted digit. This dataset can be used for tasks such as training digit extraction models, digit recognition, and sequence-based learning tasks.

• Curated by: BEN SLAMA Farah
• Language(s) (NLP): English
• License: Apache-2.0

Uses

Direct Use

The dataset is primarily intended for tasks related to:

• Digit Extraction: Extracting individual digits from numbers for training digit recognition systems.
• Sequence Learning: Useful for training models on sequence-based tasks where each digit represents a step in the sequence.
• Digit Recognition: Models focusing on recognizing or extracting specific digits from numerical inputs.

Out-of-Scope Use

This dataset is not suited for:

• Text generation or translation tasks, as it specifically deals with extracting digits from numbers.
• General language modeling tasks, as the dataset does not contain full text but rather digit-extraction sequences.

Dataset Structure

The dataset is stored in CSV format and consists of two columns:

• Description: A string description indicating the digit extraction process (e.g., "extracting 1st digit from n=123").
• Result: The extracted digit (e.g., "1").

Dataset Creation

Curation Rationale

This dataset was created to provide a structured set of digit extraction examples for machine learning tasks. It is designed to help models learn how to identify and extract digits from numbers and understand the relationships between the position of digits and the actual values.

Source Data

The source data consists of a list of 2 to 4-digit numbers from which individual digits are extracted and recorded.

Data Collection and Processing

The data is synthetically generated by:

• Iterating over numbers in a range (10 to 9999)
• Extracting each digit by position (e.g., 1st digit, 2nd digit) from the number.
• Generating a description for each extraction (e.g., "extracting 1st digit from n=123").
• The dataset is generated programmatically, ensuring consistency and accuracy in the extractions.

Who are the source data producers?

The dataset was created by BEN SLAMA Farah, CS student, who generated the dataset programmatically.

Annotations [optional]

There are no external annotations, as the dataset is generated synthetically with descriptions indicating the digit extraction process.

Personal and Sensitive Information

The dataset is free from sensitive or personal information, as it only consists of digits extracted from numbers. There are no inherent biases in the data, as the dataset is focused purely on digit extraction.

Bias, Risks, and Limitations

• Limited to numeric data: The dataset is specifically designed for digit extraction from numbers and does not cover non-numeric data or other domains.
• Synthetic nature: Since the data is synthetically generated, it may not fully capture the variety of real-world use cases for digit extraction.

Recommendations

Users should ensure that the dataset is used in contexts where digit extraction or sequence-based learning is required. It is not suitable for tasks involving unstructured text data.