TTSD: A novel dataset for Turkish Text Summarization

Abstract

Nowadays, with the increase in the amount of data on the internet, it becomes extremely important to collect salient information from these data efficiently. Gathering and comprehending essential information from such tremendous data as soon as possible is a complex and challenging process for people. Many methods have been proposed to cope with this process. Automatic text summarization systems are one of these methods in term of the fact that it facilitates obtaining the required documents by eliminating the lack of resources and time. Most of these methods are evaluated in datasets which exist in the literature. For Turkish text summarization, it is not enough datasets. In this study, we presented a novel dataset (TTSD-Turkish Text Summarization Dataset) in order to be preferred both in extractive and in abstractive methods. It is evaluated with TextRank, Lexrank, LSA-based, Luhn and Random methods by using sumy package, is compared performances using ROUGE. It is obtained that the proposed dataset has successful results in each method.