TY - JOUR AU - Lamrani Alaoui, Rokia AU - Nfaoui, El Habib PY - 2023/02/22 Y2 - 2024/03/28 TI - Cross Site Scripting Attack Detection Approach Based on LSTM Encoder-Decoder and Word Embeddings JF - International Journal of Intelligent Systems and Applications in Engineering JA - Int J Intell Syst Appl Eng VL - 11 IS - 2 SE - Research Article DO - UR - https://ijisae.org/index.php/IJISAE/article/view/2629 SP - 277 - 282 AB - <p>Web applications are the main target of Cyber Attacks. Cross-Site Scripting (XSS) is one of the most serious web attacks. Through the use of XSS, cybercriminals are able to turn trusted websites into malicious ones, resulting in extreme harm and damage to both the victims and the reputation of the website owner. According to the Open Web Application Security Project (OWASP) survey, XSS has been ranked in the top 10 web application vulnerabilities since 2017. Though its real danger, only 10 research works studied XSS attacks between 2010 and 2021 as reported recently by a systematic literature review on web attacks detection using Deep Learning. On the other hand, in many Natural Language Processing (NLP) applications, the use of word embeddings and Deep Encoder-Decoder models has considerably improved the performance of downstream NLP tasks. Thereby, in this work, we proposed a Deep Learning approach based on LSTM Encoder-Decoder and free-context word embedding for XSS attacks detection. Then, we implemented the proposed model and compared it with state-of-the-art approaches. The experimental results show that our model achieves good results; 99.08% Accuracy, 99.09% precision, and 99.08% Recall.</p> ER -