“…Furthermore, it constitutes a subset of the data being influenced by user inputs, which we call tainted data. The notion of influence between two or more data can be formally defined by the notion of causal dependency [8,19,7]. Denoting (o,t) the content of the data object o (a byte or a variable depending on the level of granularity) at time t, we may then denote (o',t') → (o,t), with t ≤ t the causal dependency of (o,t) in relation to (o',t').…”