Skip to content

This project contains codes and paperwork based on the course CSI5386 at University of Ottawa (delivered by Professor Dr. Diana Inkpen).

Notifications You must be signed in to change notification settings

kmock930/Natural-Language-Processing

Repository files navigation

Natural Language Processing Work

Assignment 1 - Corpus analysis and sentence embeddings

Assignment 1 Preview

Assignment 2 - Machine-Generated Text Detection

Assignment 2 Preview

Seminar Research - Depression Detection

Given the rising popularity of social media, there is a risk of negative impacts such as cyberbullying, causing mental health distress to some users. As a result, we dived into an exploration of depression detection with the DORIS framework proposed by Lan X., Cheng Y., Sheng L., Gao C., and Li Y. It also forms a basis for our project which aims to perform a NLP-based model targetting suicide detection.

Summary of Our Work

Summary of Models

Baseline Model

Project - Baseline Model

Fine-Tuning a Deep Learning based Transformer - DistilBERT

Project - Deep Learning based Fine-Tuning DistilBERT Model's Results

Added Custom Layers on top of Fine-Tuned DistilBERT

Project - Deep Learning based Custom Layers

Project - Deep Learning based model resulting AUC

LLM-based Model

Project - LLM-Based Model - ROC-AUC

Project - LLM-based Model - Confusion Matrix

Execution Guide

  • TMUX for idling long executions