ProbDB supports two main models, which are frequently used in uncertain applications: tuple-level and attribute-level models. In the tuple-level model, each uncertain table T has an attribute that indicates the membership probability (also called existence probability) of each tuple in T, i.e. the probability that the tuple appears in a possible world. In the attribute-level model, each attribute of each tuple can have different values each one with a probability. Formally, each tuple t_i has at least one uncertain attribute, e.g. α, and the value of α in t_i is chosen by a random variable X. Depending on the probability density function (pdf) of X, the values of α in t_i may be m values v_i,1, …, v_i,m with probabilities p_i,1, …, p_i,m respectively.