Academy
SAIL Academy aims to introduce undergraduate and associate degree students to academic research at an early stage of their education through hands-on research experience. Working alongside academic mentors, students learn how to identify research problems, explore scientific literature, critically read academic papers, design research methodologies, conduct experiments, and communicate their findings through scientific writing. The Academy also supports students in developing research proposals for the TÜBİTAK 2209-A program, guiding them from identifying a research problem and preparing a project proposal to planning and carrying out research within a defined budget.
Undergraduate Research Groups
The Undergraduate Research Groups bring students and academic mentors together to explore four research topics in cybersecurity, artificial intelligence, and computer science. Participants engage with scientific literature, develop research questions, and contribute to collaborative research activities.
Bibliometric Analysis on Turkish Computer Science
This research group analyzes publication records, titles, and abstracts of computer science articles from Türkiye. Through bibliometric analysis, it examines research trends, frequently studied topics, and how the field has developed over time.
- Student Researchers
- Zahide Gül Doğan
Muhammet Mehmet Sağlam - Academic Advisor
- Mehmet Batuhan Özdaş
LLM in Phishing URL Detection
This research group combines security assessments generated by large language models with machine learning methods to detect phishing URLs. The LLM outputs are converted into numerical features and evaluated alongside conventional URL-based approaches to determine whether they improve classification performance.
- Student Researchers
- Elif Özçelik
Sude Azra Öksüz - Academic Advisor
- Murat Osmanoğlu
LLM-Based Code Vulnerability Detection and Remediation
This group investigates how large language models identify security vulnerabilities in source code and suggest fixes. It examines whether the detected issues are valid and whether the proposed changes resolve those issues without introducing new vulnerabilities.
- Student Researchers
- Ali Kemal Çetiner
Sıla Uluyol - Academic Advisor
- Muhammed Saadetdin Kaya
LongLLMLingua-Based Intelligent RAG and Prompt Compression API
This group investigates why conventional retrieval-augmented generation (RAG) systems may miss important information located in the middle of long retrieved texts. It explores a LongLLMLingua-inspired filtering step that removes less useful content and passes a shorter, information-dense prompt to the main language model. The research compares this approach with standard RAG to evaluate information retrieval and response quality.
- Student Researchers
- Hayrunisa Yıldırım
Hayrunisa Dikme - Academic Advisor
- Salih Demir
Previously Supported Student Projects
This section presents student research projects previously accepted for funding under the TÜBİTAK 2209-A program. These projects reflect the development of student research ideas into structured studies carried out with academic guidance.
Deep Learning-Based Detection and Classification of Phishing URLs
This funded student project investigates convolutional neural network approaches for identifying phishing URLs from patterns in web addresses. It considers URL data collected from multiple sources and examines how deep-learning representations can support the classification of legitimate and potentially fraudulent links.
- Student Researchers
- Birsen Baykal
Ensar Özder
Ahmet Eray Erşahin
Hatice Kübra Ötegeçeli - Academic Advisor
- Mehmet Batuhan Özdaş
Deep Learning-Based Fraud Detection in Smart Contracts
This funded student project examines the use of convolutional and recurrent neural networks for identifying potentially fraudulent behavior in blockchain smart contracts. Its research focuses on suspicious patterns in contract-related information and the suitability of different deep-learning approaches for detecting security-relevant anomalies.
- Student Researchers
- Ardacan Bodur
Emine Kurtay
Zeynep Demir
Habil Ernekal - Academic Advisor
- Mehmet Batuhan Özdaş
Student Research Projects in Preparation
This section presents student research proposals being developed for the TÜBİTAK 2209-A program, including projects undertaken through SGM261. Students identify research problems, review related work, and plan methods and experiments with academic guidance.
Model Robustness in Malicious URL Detection
This project evaluates how malicious URL detection models respond to unfamiliar URL patterns and altered inputs. It examines whether changes to the structure of a URL affect classification decisions and compares model performance under different test conditions.
- Student Researchers
- Muhammed Ali Boğaz
- Academic Advisor
- Muhammed Saadetdin Kaya
Text-Based Detection of Phishing Emails
This project investigates phishing detection through the text of email messages. It examines language-based indicators of fraudulent communication and evaluates classification methods for distinguishing phishing emails from legitimate messages.
- Student Researchers
- Durukan Mete Duman
Mehmet Erhan Güleç - Academic Advisor
- Mehmet Batuhan Özdaş
Detection of Algorithmically Generated Domain Names
This project studies how automatically generated domain names can be distinguished from ordinary domain names. It focuses on patterns in domain name strings and the evaluation of methods for identifying potentially suspicious algorithmically generated domains.
- Student Researchers
- Özge Can
- Academic Advisor
- Muhammed Saadetdin Kaya
Network Intrusion Detection and False Positive Analysis
This project examines methods for detecting potential intrusions in network traffic, with particular attention to false positive alerts. It compares how detection approaches distinguish attacks from legitimate activity and evaluates the balance between detection performance and false alarms.
- Student Researchers
- Emirhan Öztaş
Osman Demir - Academic Advisor
- Mehmet Batuhan Özdaş
Indicator of Compromise Extraction from Threat Intelligence Texts
This project explores the identification of indicators of compromise (IoCs) within unstructured threat intelligence texts. It examines how relevant security information, such as suspicious addresses, domains, or file indicators, can be extracted and organized for further analysis.
- Student Researchers
- Bedirhan Gökdemir
Elmas Sıla Özkan - Academic Advisor
- Muhammed Saadetdin Kaya
Vulnerability Prioritization from CVE Descriptions
This project investigates how information in Common Vulnerabilities and Exposures (CVE) descriptions can support vulnerability prioritization. It examines textual characteristics of vulnerability reports and evaluates approaches for organizing vulnerabilities according to their potential security significance.
- Student Researchers
- Recep Burak Akar
- Academic Advisor
- Murat Osmanoğlu
Human-in-the-Loop AI Agent for Security Operations
This project explores an AI-assisted agent that helps with selected security operations tasks while keeping a human analyst involved in decisions. It examines workflows in which the agent analyzes information and proposes actions that require human review or approval.
- Student Researchers
- Zekeriya Fırat Bozkırlı
Ezelnaz Erdem - Academic Advisor
- Mehmet Batuhan Özdaş
Suspicious Behavior Detection in Failed Login Logs
This project analyzes failed login records to identify patterns that may indicate suspicious authentication activity. It examines how repeated failures and related log characteristics can help distinguish possible attacks from normal user mistakes.
- Student Researchers
- Burak Gündoğmuş
- Academic Advisor
- Muhammed Saadetdin Kaya
Suspicious Request Analysis in Web Server Access Logs
This project investigates potentially malicious requests recorded in web server access logs. It examines request patterns and recurring behavior to identify unusual activity that may indicate attempts to probe or exploit a web application.
- Student Researchers
- Yavuz Selim Bozdemir
Görkem Arslan - Academic Advisor
- Muhammed Saadetdin Kaya
Mapping Security Incident Descriptions to MITRE ATT&CK
This project examines the automatic mapping of natural-language security incident descriptions to relevant MITRE ATT&CK techniques. It focuses on interpreting incident narratives and identifying the attack techniques they describe to support more consistent threat analysis.
- Student Researchers
- İrem Genç
Tuğçe Çapacı - Academic Advisor
- Mehmet Batuhan Özdaş
Rule-Based and Machine Learning Detection of Spatial and Temporal Inconsistencies in ADS-B Flight Data
This project investigates anomalies in aircraft position, speed, and altitude data transmitted through ADS-B. It plans to collect real flight data using a single low-cost receiver and generate simulated spoofing, ghost aircraft, and replay scenarios in an offline environment. Rule-based checks, Isolation Forest, and a hybrid detection approach will be compared to assess attack detection and false alarms caused by signal gaps or missing messages.
- Student Researchers
- Ceren Nur Savaş
Fadıl Yeren - Academic Advisor
- Mehmet Batuhan Özdaş