SecurityAttackTypeDataset

The purpose of these datasets is to recognize the importance of the relationships between complexity and vulnerability metrics used in identifying software security vulnerabilities and their security attack types at the function level. Using these datasets as input to machine learning and deep learning models can be developed top security vulnerability detection models with high accuracy. As you can see, this is a repository of datasets used to detect security vulnerabilities and types of vulnerabilities at the function level. The difference between these datasets and previous datasets is that they are prepared separately for each type of security attack. In each set of security attacks, the number of vulnerable functions is balanced by the number of non-vulnerable functions that are extracted from some real-world projects. for example, the zip file related to the DoS attack, shows you eleven datasets. in each dataset of the first ten datasets, you can see vulnerable rows related to DoS attack and non-vulnerable rows together as balanced. note that non-vulnerable rows were extracted distributedly from some real-world projects. Each dataset has complexity metrics (C1,.., C4) and vulnerability metrics (V1,..., V11). Using existing algorithms and methods in machine learning and deep learning and using these datasets as model input, we were able to create models in vulnerability detection and security attack type that have very high accuracy. We were also able to identify the relationships between the metrics introduced above. So we can find out which of these criteria is really more effective in diagnosing vulnerability than other criteria.