Abstract: When the attribute set α of the common set X changes dynamically, the unknown data information is hidden in X. According to the situation, the concepts of internal P-unknown data set, outer P-unknown data set and P-unknown data sets are proposed by using P-sets(X^(-overF),X^F)generated by X and the structure of ∧-type big data, and their numerical characteristics are given. The concepts of P-dependence degree and P-filtering degree of unknown data sets are defined, and the relationships between P-dependence degree and P-filtering degree are discussed. The conditions for filtering and identifying P-unknown data sets are analyzed by the filtering degree, and the filtering-separation and identification principle, the filtering surplus-separation and identification principle of P-unknown data sets are given, the filtering-separation and identification theorems and the filtering surplus-separation and identification theorems of P-unknown data sets are obtained. Finally, the application of filtering and identifying internal P-unknown data set is given.

Key words: P-set, P-unknown data set, P-dependence degree, P-filtering degree, filtering-separation