v1.9.9

This source code was used in the bachelor thesis “Public Sentiment Analysis of Mental Disorder based on Twitter Texts using Support Vector Machine”.

This research is an experimental research focused on modeling sentiment analysis systems built on the support vector machine (SVM) method. The expected output of the experiment is to obtain a model (combination of SVM kernel and feature extraction technique) with the best performance as measured by the confusion matrix.

In general, the source code contains the Python package, Jupyter notebook (interactive Python computing document), and datasets. The details are as below.

Module:

• Scrape
• Annotation
• Analysis
• Helper
• Slang template
• Slang dictionary
• Antonym template
• Antonym dictionary
• Preprocessing
• Encoding
• Feature extraction
• Feature inspection
• SVM classification
• SVM plot
• Cross validation
• Model tuning
• Model testing

Notebook:

• Main (ta.ipynb)
• Training and testing
• Plotting
• Revised training and testing (after thesis trial)
• Revised plotting (after thesis trial)

Dataset:

• Main (tweets)
• Indonesian slang dictionary
• Indonesian antonym dictionary
• Data output from model training