Türkçe Akademik Yayınlar İçin Yapısal Öz Çıkarım Sistemi
Özkan Çelik, Ayşe Esra
xmlui.mirage2.itemSummaryView.MetaDataShow full item record
With the widespread use of information technologies in all areas of life, the long-lasting problem of researchers' inability to read the full texts of articles has gained a different dimension. Researchers are also tackling new challenges while studying the massive amount of information they access in the electronic media due to the nature of electronic versions of texts, the online media, and the digital screens. Abstracts are the most important tools that will enable potential readers to select the relevant full texts from this huge stack of electronic information. However, whereas their primary aim is to facilitate the reading of the full texts, in most cases the abstracts are produced with more readability issues and structural deficiencies in their contents when compared the actual texts. In this dissertation, ways of contributing to the readers’ experience of meeting scientific articles in today’s conditions, where technology constantly transforms people’s interaction with information, were investigated. Accordingly, the study focuses on the contribution of the use of structural methods in obtaining the abstracts of scientific articles through automatic methods. The system developed in the study was designed to operate with the principle of choosing representative sentences from all structural parts of the full texts of the articles. The automatically structured abstracts obtained are superior to the classical version of outputs of the system. The calculated legibility values show that all the outputs are significantly easier to read than those produced by authors. The summarization system, which was developed within the scope of the study, is independent from the data set. The summarization system can also be used for other Turkish document groups. It is believed that the results of this study and the data set, which was extracted from a total of 421 articles published in the field of Turkish librarianship and information science, can be used in other studies as well. It is considered that this structured abstract preparation system can be used to produce better readable automatic structural abstracts for the corpus produced with classical abstracts, and can also contribute to the scientific communication process as an auxiliary tool for authors and editors.