Frequent Pattern Mining

Years and Authors of Summarized Original Work

• 2004; Uno, Kiyomi, Arimura

Problem Definition

Pattern mining is a fundamental problem in data mining. The problem is to find all the patterns appearing in the given database frequently. For a set E = { 1, …, n} of items, an itemset (also called a pattern) is a subset of E. Let $\mathcal{D}$ be a given database composed of transactions R₁, …, R _m , R _i  ⊆ E. For an itemset P, an occurrence of P is a transaction of $\mathcal{D}$ such that P ⊆ R, and the occurrence set Occ(P) is the set of occurrences of P. The frequency of P, also called support, is | Occ(P) | and denoted by frq(P). For a given constant σ called minimum support, an itemset P is frequent if frq(P) ≥ σ. For given a database and a minimum support, frequent itemset mining is the problem of enumerating all frequent itemsets in $\mathcal{D}$.

Key Results

The enumeration is solved in output polynomial time [1], and the space complexity is input polynomial [5]. Many algorithms...