A Novel Hybrid Architecture for Distributed Knowledge Discovery in Scientific Environment

摘要

Managing data and knowledge sharing in scientific environment is a great challenge, especially in today's big data environment. Traditional monolithic Business Intelligence (BI) systems are difficult to scale up to the massive data sets distributed across a large number of sites. Targeting this challenge, the data mining research has to overcome two major obstacles; mmimg in a distributed environment, and mining on the fly in order to pursue knowledge discovery. Unfortunately, due to highly domain dependency and monolithic deployment, the integration of such large-scale mining on big data is proof to be a tall order. In this paper, we propose a flexible mining system with multi-tier architecture to alleviate these challenges. The system meets the diverse needs of scientific data mining and knowledge sharing. Each service represents the result of an orchestration of reusable building blocks as well as predefined tasks. Its main advantage is that it can deliver BI process as Software-as-a-Service and supports adaptive data mining processes.