- El-Cezeri
- Vol: 9 Issue: 4 Özel Sayı
- Makine Öğrenmesi Algoritmaları ile Uçtan Uca Yazar Tanıma Uygulaması Geliştirme
Makine Öğrenmesi Algoritmaları ile Uçtan Uca Yazar Tanıma Uygulaması Geliştirme
Authors : Ilayda Erdoğan, Merve Güllü, Hüseyin Polat
Pages : 1303-1314
Doi:10.31202/ecjse.1134698
View : 15 | Download : 7
Publication Date : 2022-12-31
Article Type : Research
Abstract :Abstract: The problem of unidentified texts, which has been going on for centuries, has increased considerably with the beginning of the internet age. The biggest reason for this situation is that very high percentage of data on the internet is composed of unstructured data, and a large part of this unstructured data is composed of unclassified texts with uncertain authors. The use of machine learning methods in classification processes in recent years has brought a new perspective to authorship identification problems. In this study, an end-to-end application with a web-based interface was developed for the authorship identification problem using machine learning methods. First of all, a corpus containing 46715 text data was created from the columns of 37 authors. Features were extracted from this corpus using the TF-IDF method and a dataset was obtained. Then the dataset is trained and tested with Support Vector Machines (SVM), NB (NB) and Random Forest (RF) machine learning algorithms. As a result of the test, SVM was the best performing classifier model with 90% accuracy. A web interface was developed for the obtained SVM model by using Flask, one of the libraries of the Python programming language. Then, the application has been converted into a Docker container to run it in a stable and distribution-friendly state. As a result, an end-to-end authorship identification application has been made to deploy available directly to the end user. The creation of such a web-based application with the support of machine learning has made the authorship identification study more meaningful and usable.Keywords : Yazar tanıma, Destek Vektör Makineleri, TF-IDF, Docker