Grammar-aware phrase dataset generated using a novel python package

Area of Science:

• Computational Linguistics
• Natural Language Processing
• African Languages

Background:

• Manual dataset preparation is labor-intensive and prone to errors.
• Existing web scraping methods for data acquisition also yield data inaccuracies.
• Developing automated tools is crucial for efficient linguistic data processing.

Purpose of the Study:

• To introduce "Oromo-grammar," a novel Python package for automated Oromo language dataset generation.
• To overcome the limitations of manual and web-scraped data preparation.
• To create a grammar-rich dataset applicable to NLP and linguistic studies.

Main Methods:

• The "Oromo-grammar" package accepts raw text files as input.
• It extracts root verbs and generates corresponding verb stems.
• The algorithm synthesizes grammatical phrases with affixations and pronouns, indicating grammatical features like number, gender, and case.

Main Results:

• A novel Python package, "Oromo-grammar," was developed.
• The package successfully generates a grammar-rich dataset of Oromo phrases.
• The generated dataset includes grammatical information such as number, gender, and case.

Conclusions:

• The "Oromo-grammar" package offers an efficient and reproducible method for Oromo language dataset creation.
• The generated dataset supports advanced Natural Language Processing (NLP) applications, including machine translation and grammar checking.
• This approach can be adapted for other languages with systematic analysis and minor modifications.