Filtering Turkish Spam Using LSTM from Deep Learning Techniques

Eryilmaz E. E., Şahin D. Ö., Kılıç E.

8th International Symposium on Digital Forensics and Security, ISDFS 2020, Beirut, Lebanon, 1 - 02 June 2020 identifier

  • Publication Type: Conference Paper / Full Text
  • Doi Number: 10.1109/isdfs49300.2020.9116440
  • City: Beirut
  • Country: Lebanon
  • Keywords: artificial intelligence, deep learning, Keras library, LSTM, machine learning, spam detection, Turkish spam filtering
  • Ondokuz Mayıs University Affiliated: Yes

Abstract

E-mails are used effectively by people or communities who want to do propaganda, advertisement, and phishing because of their ease of use and low cost. People or communities who want to achieve their goals send unnecessary and spam to the e-mail accounts they never knew. These mails cause serious financial and moral damages to internet users and also engage in internet traffic. Unsolicited e-mails (spam) are a method sent to the recipient without their consent and generally for malicious or promotional purposes. In this study, spam was detected with Keras deep learning library on the Turkish dataset. Turkish email dataset contains 800 e-mails, half of which are spam e-mails. With the deep learning algorithm long short term memory (LSTM), a 100% accuracy rate has been achieved in the Turkish e-mail dataset.