Sentiment Analysis for Software Engineering Domain in Turkish

Abstract

The focus of this study is to provide a model to be used for the identification of sentiments of comments abouteducation and profession life of software engineering in social media and microblogging sites. Such a pre-trainedmodel can be useful to evaluate students’ and software engineers’ feedbacks about software engineering. Thisproblem is considered as a supervised text classification problem, which thereby requires a dataset for the trainingprocess. To do so, a survey is conducted among students of a software engineering department. In the classificationphase, we represent the corpus by using conventional and word-embedding text representation schemes and yieldaccuracy, recall and precision results by using conventional supervised machine learning classifiers and wellknown deep learning architectures. In the experimental analysis, first we focus on achieving classification resultsby using three conventional text representation schemes and three N-gram models in conjunction with fiveclassifiers (i.e., naïve bayes, k-nearest neighbor algorithm, support vector machines, random forest and logisticregression). In addition, we evaluate the performances of three ensemble learners and three deep learningarchitectures (i.e. convolutional neural network, recurrent neural network, and long short-term memory). Theempirical results indicate that deep learning architectures outperform conventional supervised machine learningclassifiers and ensemble learners.