COMMAX 2020 UiTM Kampus Kuala Terengganu

Companies would surely want to know how the public feels about them and how they are perceived as in the eyes of their customers or clients. This project aims to provide that exact insight, specifically for Muslim vendors, as a classifier for BMF messages on Twitter was developed by using deep convolutional neural network (DCNN) algorithm. DCNN is a popular machine learning algorithm that mimics the functionality of an actual human neural network. It is a type of artificial neural network or ANN. By using machine learning algorithms such as DCNN, the public opinion can be extracted and understood. This is useful information for the Muslim market which can reveal whether the public is supporting BMF or otherwise. This project focuses on English language reviews regarding BMF on the Twitter social media platform. Tweets were downloaded directly from Twitter by using the Twitter4J library which allows Java applications to access the Twitter API. Each tweet is stored in an individual notepad file, which then gets passed to a python code to be cleaned. The processes involved in cleaning include lower-casing, removing stop words, and removing low significant words. The accuracy of the model is currently low due to certain factors, but more improvements can be done in the future. The model can yield only 55% accuracy for now. More research is required to finetune the model which will improve the classification accuracy. This includes collecting more data, collecting precise data that relates to the area of interest, and finding the most optimum values for the parameters of the model.