Natural Language Processing for Decolonizing Harm Reduction Literature

Duration: 9 weeks

Develop an NLP model to analyze existing harm reduction literature and identify colonial language, perspectives, or biases. The system could suggest alternative phrasings and highlight areas where Indigenous and diverse voices are underrepresented, assisting in the decolonization of harm reduction materials.

Final Deliverables:

A final report including:

• Detailed analysis of the harm reduction literature dataset
• Description of the problem solved (identifying and addressing colonial language in harm reduction literature)
• Comprehensive explanation of methodologies and approaches used
• Outcomes and results of the NLP models
• Recommended next steps for implementation and further development

Source materials including:

• All code used in data preprocessing, model development, and evaluation
• Jupyter notebooks or similar workbooks documenting the entire process
• Trained model files and any necessary dependencies

A presentation summarizing the project, key findings, and potential applications for the J Healthcare Initiative

Project: Natural Language Processing for Decolonizing Harm Reduction Literature

Duration: 9 weeks

Project Overview:

Develop a Natural Language Processing (NLP) model to analyze existing harm reduction literature, identify colonial language and biases, and suggest decolonized alternatives. This project aims to support the J Healthcare Initiative's mission of decolonizing harm reduction.

By the end of the project, students should demonstrate:

1. Understanding of the available dataset:

• Comprehensive analysis of the harm reduction literature corpus
• Identification of key themes, terminology, and linguistic patterns in the dataset
• Recognition of potential biases and colonial perspectives in the literature

1. Understanding of the latest AI/ML techniques:

• Proficiency in current NLP methodologies, including transformer models like BERT or GPT
• Knowledge of sentiment analysis, named entity recognition, and text classification techniques
• Familiarity with bias detection and mitigation strategies in NLP

1. Identification of ways in which AI/ML can be applied to the J Healthcare Initiative:

• Propose methods for automated detection of colonial language and biases
• Suggest AI-driven approaches for generating decolonized alternative phrasings
• Outline potential applications of the NLP model in reviewing and improving harm reduction materials

Bonus steps:

• Provide multiple versions of potential models, such as:
• a) A binary classification model for identifying colonial vs. decolonized language
• b) A multi-class model for categorizing different types of biases or colonial perspectives
• c) A generative model for suggesting decolonized alternatives to identified colonial language

Methodology:

Week 1-2: Data Collection and Preprocessing

• Gather a diverse corpus of harm reduction literature
• Preprocess the text data (tokenization, cleaning, etc.)
• Perform exploratory data analysis to understand the dataset characteristics

Week 3-4: Model Development (Version 1)

• Implement a basic NLP model for identifying colonial language
• Train and evaluate the model using appropriate metrics

Week 5-6: Advanced Model Development

• Develop more sophisticated models, potentially including:
• Fine-tuned transformer models for nuanced language understanding
• Ensemble methods combining multiple NLP techniques
• Implement bias detection and mitigation strategies

Week 7-8: Model Evaluation and Refinement

• Conduct thorough testing of the models with diverse test sets
• Refine models based on performance and feedback from domain experts
• Develop a system for suggesting decolonized alternatives to identified colonial language

Week 9: Documentation and Presentation

• Prepare the final report and source materials
• Develop recommendations for practical application and next steps

Students will connect directly with us for mentorship throughout the project. We will be able to provide answers to questions such as:

• Current industry standard approaches to AI / ML
• Input on choices, problems or anything else the students might encounter.
• One to one meetings for feedbacks