Programming parallel apriori algorithms for mining association rules

Many parallel programming languages provide a set of low-level parallel constructs which lead to programs that are difficult to design, implement, debug, and maintain. In this paper, we present a parallel programming paradigm using a sequential programming language that encapsulates low-level details of distributed and concurrent programming techniques and makes the details transparent to programmers. This paradigm allows programmers to program on several different platforms in a sequential language that they are most familiar with. In addition, the paradigm provides high-level language constructs for programmers to implement various distributed and parallel algorithms. The programming paradigm is illustrated with two parallel Apriori algorithms in data mining.