M. Muyeba and R. Mamadapali (UK)
Rule, query, distributed, induction, filtering, performance
Inductive algorithms produce rules that are used to evaluate and interpret their interestingness while some use rule querying as a post data mining technique for query optimisation purposes. The need for interactive and user query tools demand a data mining query language. The problem here addresses post rule based querying in a distributed environment. The generalization method of Attribute-Oriented Induction (AOI) with key-preservation (AOI-KP) is used to associate every rule with a collection of tuple keys at each host processor and utilizes a star schema database design. The key-preservation aspect seems to address the information loss problem in AOI and may help to perform efficient data queries using tuple keys. An SQL-like data and rule query operator that utilises data mining primitives in a distributed environment is presented with appropriate rule filtering mechanisms. Initial performance results indicate good scalability and speed-up in a distributed environment.
Important Links:
Go Back
