A Machine Learning Model for Sentiment Classification of Mobile Products Data
Rebhi S. Baraka, Yasmeen H. Jaber · 2023
Negative or positive opinions regarding a product affect the purchasing decisions of potential customers and hence the producing company. Based on these opinions, there is a need to determine the features affecting the product and its producer. In this paper, we present a machine learning model we have built for sentiment analysis of mobile product sentiment data coming from Twitter. It aims to categorize sentiments for a particular product of a particular mobile phone company. The model applies three classification algorithms; Naive Bayes (NB), Logistic Regression (LR), and Support Vector Machine (SVM) using two types of features, namely discrete features and continues features. We have compared their results using two types of classification models, supervised and semi-supervised with several test cases. With Fl-score of 80.26% for supervised and 87.64% for semi-supervised, the semi-supervised model outperforms the supervised model. Further research on continuous features is needed in the future as it does not require the use of a feature engineering process. This includes the use of word embeddings which perform well in related tasks.