Code and Dataset for "Invisible Labour, Categorised Threat: A Computational Analysis of How Indian English-Language Media Frames Interstate Migrants in Kerala"

This repository contains the code and derived dataset supporting the paper "Invisible Labour, Categorised Threat: A Computational Analysis of How Indian English-Language Media Frames Interstate Migrants in Kerala." The study examines how interstate migrant workers are represented in English-language Indian media through computational analysis of 966 news articles from 180 media sources, published between 2020 and 2026. The analysis pipeline combines Latent Dirichlet Allocation (LDA) topic modelling, transformer-based sentiment classification (cardiffnlp/twitter-roberta-base-sentiment-latest) and emotion classification (emotion-english-distilroberta-base), context-filtered lexicon analysis of othering/inclusive language, and inferential statistical testing.