AI Application for Media Accountability in Drug User Representation
The J Healthcare Initiative is a non-profit organization that does projects to tackle the overdose crisis. In this project we address the pervasive issue of hate rhetoric, stigmatizing language, and misinformation in media coverage related to drug users and harm reduction through machine learning/AI. This project seeks to create a comprehensive analysis of current media practices and develop guidelines for responsible reporting. The goal is to ensure that media outlets are held accountable for their portrayal of drug users, promoting a more accurate and compassionate narrative. By doing so, the project will contribute to reducing stigma and fostering a more informed public discourse. Objective: Develop a basic tool to analyze sentiment and language use in media articles about drug use and harm reduction. Suggested timeline Python Script for Media Analysis (25 hours) Develop a script that: Scrapes articles from 2-3 selected news websites Performs basic sentiment analysis Identifies key terms related to drug use and harm reduction Flags potentially stigmatizing language Small Dataset of Analyzed Articles (10 hours) Collect and analyze 50-100 recent articles Store results in a structured format (e.g., CSV or JSON) Summary Report (15 hours) Brief overview of the project and methodology Key findings from the analysis 2-3 data visualizations (e.g., sentiment distribution, word clouds) Short list of common stigmatizing terms found Basic Web Interface (10 hours) Simple Flask or Streamlit app that: Allows users to input an article URL or text Runs the analysis script on the input Displays basic results (sentiment score, flagged terms) Key objectives include: - Analyzing existing media content for instances of hate rhetoric and misinformation. - Identifying patterns and common issues in media representation of drug users. - Developing a set of guidelines for responsible and accurate reporting. - Proposing strategies for media accountability and public awareness.
