Akademik Veri Yönetim Sistemi
Dokuz Eylül Üniversitesi Mühendislik Fakültesi Fen ve Mühendislik Dergisi, cilt.24, sa.72, ss.825-834, 2022 (Hakemli Dergi)
Defective expression is a grammatical term that refers to both semantic and morphologic ambiguities in Turkish sentences. In earlier studies, Natural Language Processing (NLP) techniques have been used by constructing rule-based language-specific models. However, despite less demanding annotations requirements and ease of incorporating external knowledge, rule-based systems have some significant obstacles in terms of processing efficiency. Deep learning techniques such as long short-term memory (LSTM) or convolutional neural network (CNN) have made significant advances in recent years, which led to an unprecedented boost in NLP applications in terms of performance. In this study, a hybrid approach of LSTM and CNN (C-LSTM) for detecting defective expressions in addition to traditional machine learning classifiers such as support vector machine (SVM) and random forest (RF) to compare the results in terms of accuracy are proposed. The proposed hybrid approach achieved higher accuracy than the existing deep neural models of CNN and LSTM, in addition to the traditional classifiers of SVM and random forest. This study shows that deep neural approaches come into prominence for text classification compared to traditional classifiers.