Utilizing Machine Learning with Unique Pentaplet Data Structure to Enhance Data Integrity

Abstract

Data protection in databases is critical for any organization, as unauthorized access or manipulation can have severe negative consequences. Intrusion detection systems are essential for keeping databases secure. Advancements in technology will lead to significant changes in the medical field, improving healthcare services through real-time information sharing. However, reliability and consistency still need to be solved. Safeguards against cyber-attacks are necessary due to the risk of unauthorized access to sensitive information and potential data corruption. Disruptions to data items can propagate throughout the database, making it crucial to reverse fraudulent transactions without delay, especially in the healthcare industry, where real-time data access is vital. This research presents a role-based access control architecture for an anomaly detection technique. Additionally, the Structured Query Language (SQL) queries are stored in a new data structure called Pentaplet. These pentaplets allow us to maintain the correlation between SQL statements within the same transaction by employing the transaction-log entry information, thereby increasing detection accuracy, particularly for individuals within the company exhibiting unusual behavior. To identify anomalous queries, this system employs a supervised machine learning technique called Support Vector Machine (SVM). According to experimental findings, the proposed model performed well in terms of detection accuracy, achieving 99.92% through SVM with One Hot Encoding and Principal Component Analysis (PCA).