Motivated by applications in online privacy, user authentication and profiling, we discuss the complexity of various problems generalized from the classic 0-1 knapsack problem. In our scenarios, we assume the existence of a scoring function that evaluates the confidence in the personal online profile or authenticity of an individual based on a subset of acquired credentials and facts about an individual and show how the specific properties of that function can affect the computational complexity of the problem, providing both NP-completeness proofs under certain conditions as well as pseudo-polynomial-time solutions under others.