Efficient Turkish tweet classification system for crisis response

Abstract

This paper presents a convolutional neural networks Turkish tweet classification system for crisis response. This system has the ability to classify the present information before or during any crisis. In addition, a preprocessing model was also implemented and integrated as a part of the developed system. This paper presents the first ever Turkish tweet dataset for crisis response, which can be widely used and improve similar studies. This dataset has been carefully preprocessed, annotated, and well organized. It is suitable to be used by all the well-known natural language processing tools. Extensive experimental work, using our produced Turkish tweet dataset and the English dataset ("socialmediadisaster-tweets-relevent"), has been performed to illustrate the performance of the developed approach. In addition, vector space model (VSM) techniques were studied to find out the most suitable technique that can be used for the Turkish language. Overall, the developed approach has achieved a quite good performance, robustness, and stability when processing both Turkish and English languages. Our experiments also compare the performance with some stateof-the-art English language systems, such as "CREES" and "deep multimodal".