Bangla_MER: A unique dataset for Bangla mathematical entity recognition.

Area of Science:

• Natural Language Processing
• Computational Linguistics
• Artificial Intelligence

Background:

• Mathematical entity recognition is crucial for computational understanding of mathematical content.
• Existing resources for Bangla mathematical entity recognition are scarce, hindering research and development.
• A dedicated dataset is needed to advance machine learning models for Bangla mathematical expressions.

Purpose of the Study:

• To introduce the first comprehensive dataset for Bangla mathematical entity recognition.
• To facilitate research in identifying mathematical operators, terms, and operands in the Bangla language.
• To support the development of advanced machine learning and deep learning models for Bangla mathematics.

Main Methods:

• Compilation of a novel dataset comprising 13,717 observations.
• Each record includes a mathematical statement, its type, and the identified mathematical entity.
• Dataset structured in raw format and provided as a CSV file with 'text', 'math entity', and 'label' columns.

Main Results:

• Creation of a state-of-the-art Bangla mathematical entity dataset.
• The dataset enables recognition of operators, mathematical terms (e.g., complex numbers), and numerical operands.
• The dataset is amenable to research on combined entity recognition with minor modifications.

Conclusions:

• The presented dataset is a significant contribution to Bangla Natural Language Processing and AI research.
• It provides a foundation for developing and evaluating machine learning models for mathematical entity recognition in Bangla.
• Researchers can leverage this dataset for diverse explorations in deep learning and machine learning applied to mathematical language.