README

This repo serves as the code implementation of the project assigned in the module Applied Machine Learning Systems II ELEC0135.

Description

In this project, we are required to solve two tasks upon Twitter sentiment analysis in the field of natural language processing. The first task (A) deals with a three-point scale tweet polarity classification problem, namely to classify whether the sentiment of the tweet is positive, negative or neutral, while the second task (B) works on a two-point scale topic-based message polarity classification issue, aiming to distinguish the positive and negative sentiments of tweets towards the topic. In this paper, a method of Recurrent Neural Network (RNN), or more precisely Long Short-Term Memory (LSTM) is applied to tackle these two tasks, along with a complete text preprocessing pipeline and word embedding approaches. The combination of these techniques has brought us a high classification accuracy of 62.9% for task A and 86.5% for task B. Need to mention that these two tasks are originally the Competition Task 4 held by SemEval 2017 where our results rank 13rd/38 and 3rd/23 on the list although we are not officially participating.

Results

Task	Model	Train Acc	Test Acc	Epoch Count	Elapsed Time
A	Bi-LSTM	72.8%	62.9%	17	00:27:50
B	Bi-LSTM	95.1%	86.5%	18	00:15:43

Requirements

• Python 3.7+ (Python 3.7.10 is recommended which is the version used in the development.)
• macOS or Windows
• Required modules
  • check requirement.txt to ensure all modules included have been installed
  • or run pip3 install -r requirements.txt in terminal to install with ease

Usage

1. Switch to the root directory of the repo
2. Run python3 main.py in terminal to start the main program
3. Monitor the running process in base_log.log if you like (logging level is set to INFO by default, feel free to alter)

Structure

├── A                      # the folder of Task A
│   ├── A_lstm_model.h5    # trained model for Task A
│   └── A_sentiment_classification.py                     # scripts for Task A
├── B                      # the folder of Task B
│   ├── B_lstm_model.h5    # trained model for Task B
│   └── B_topic_based_sentiment_classification.py         # scripts for Task B
├── Datasets               # the folder of datasets
│   ├── 4A-English
│   └── 4B-English
├── README.md               # this file
├── helper                  # the folder of some necessary files
│   ├── a_learning curve.png  # learning curve for Task A
│   ├── b_learning curve.png  # learning curve for Task B
│   └── base_log.log        # the main log, generated once the main program srarts
├── main.py                 # the entrance to main program
└── requirements.txt        # the list of required libraries

Having problems?

If you run into problems, please either file a github issue or send an email to [email protected].