Bapynshngain/FEM-Khasi-News-Monolingual-Corpus

Khasi Monolingual News Corpus (740K)

Project Attribution & Collaboration

This dataset was collected and curated as part of the research project titled "Financial Empowerment in Meghalaya: AI-Powered Multilingual E-Marketplace for Tribes."

This project is a collaborative research initiative conducted by:

• National Law University (NLU) Meghalaya
• Indian Institute of Information Technology (IIIT) Guwahati

Contributors: This dataset is the result of a joint effort by the project team. Key contributors to the data collection, cleaning, and curation process include:

• Dr. Basil N. Darlong Diengdoh (Principal Investigator)
• Dr. Kaustuv Nag (Co-Principal Investigator)
• Dr. Dipankar Kundu (Co-Principal Investigator)
• Bapynshngainlang Nongkynrih (Junior Research Fellow)
• Ahlad Pataparla (Project Intern)

(Note: This dataset was previously uploaded independently to the Hub under the username 'user10383'. This repository serves as the official, fully attributed version representing the collaborative work of the NLU Meghalaya and IIIT Guwahati research teams.)

Data Instances

kha: A string containing a Khasi sentence.

Source Data

The data was primarily scraped and compiled from digital archives of Khasi language news publications and other publicly accessible Khasi text sources.

Data Cleaning

The corpus has undergone cleaning to remove:

• HTML tags and formatting artifacts.

• Non-Khasi text blocks (where identifiable).

• Duplicate sentences.

Known Limitations

• Domain Bias: As the data is heavily sourced from news articles, models trained primarily on this corpus may exhibit a formal or journalistic tone and might lack representation of conversational, colloquial, or dialectal Khasi.

• Transcription Errors: Web-scraped data may contain inherent typographical errors or inconsistent spelling conventions that exist in the source material.

Citation Information

If you use this dataset in your research, please attribute the "Financial Empowerment in Meghalaya: AI-Powered Multilingual E-Marketplace for Tribes" project team (NLU Meghalaya & IIIT Guwahati).

Dataset Structure

The dataset is formatted as a CSV file consisting of a single column of text containing Khasi monolingual sentences, primarily sourced from regional news outlets.

Data Instances

A typical instance in the CSV file looks like this:

kha
"Ka sorkar jylla ka la pynbna ba kan plie biang ia ki skul naduh ka taiew ban wan."