Deep Learning Group Course Work - 2023

Yelp Review Data Analysis with Deep Learning Models - Sentiment Analysis

Group Members

• Lakshani Nissanka - 20210570
• Suvini Viduneth - 20210569
• Kavindya Koralegei - 20210575

Overview

Implementation of a Supervised and Unsupervised Models to find the Optimal model that can be used for the sentiment analysis

Models Using:

• Convolutional Neural Networks (CNN)
• BERT with K-Means clustering

Dataset

• Link to the Yelp Reviews dataset.
• We are using the - The json file "yelp_academic_dataset_review.json" for our purpose.

Project Structure

project-root/
│
├── data/
│   ├── yelp_dataset.json       # Downloaded Yelp dataset
│
├── models/
│   ├── supervised_model/       # Code and files related to the supervised model
│   ├── unsupervised_model/     # Code and files related to the unsupervised model
│
├── notebooks/
│   ├── Preprocessing.ipynb  # Notebook for the supervised model
│   ├── Supervised_Model_CW2.ipynb # Notebook for the supervised model
│   ├── Unsupervised_Model_CW2.ipynb # Notebook for the unsupervised model
│
├── README.md                   # This README file

Data Preprocessing

-You can reference the Preprocessing.ipynb notebook for details.

Supervised Model

-We have used CNN model as the Supervised Model for this task. -Reference the Supervised_Model_CW2.ipynb notebook for code and results.

Unsupervised Model

-We have used BERT along with K-Means clustering Model for this task. -We used a pretrained BERT model from hugging face to implement the BERT model. -Reference the Unsupervised_Model_CW2.ipynb notebook for code and results.