DAG Based Data Generation with Functional Dependencies Consistency

For data generation problems with functional dependency (FD) consistency, directed acyclic graph (DAG) was used to model FDs set, an algorithm of tuple generation with single FD (TGSFD) was proposed to generate data consistent with single FD, an attributes sorting algorithm was proposed to solve the data generation problems with multi FDs. In order to utilize pipelining technique to improve the efficiency of data generation, a concept of minimal independent attributes subset (MIAS) was proposed and the attributes set partitioning algorithm was given. Experiments results show that TGSFD and attributes sorting algorithm can guarantee the FD consistency of generated data, while MIAS and pipeline technique can improve the efficiency of data generation.