Research Challenges for Data Mining (1)

• Scalability
• efficient and sufficient sampling schemes
• in-memory vs. disk-based data processing
• choice of right subset of techniques to span most tasks
• interfaces to large warehouses, use of metadata to optimize access
• client-server issues, where to perform the processing (where and when to mine)
• exploiting parallelism, distributed computing over a network of workstations

General systems challenge: what will a system that will enable exploration, visualization, analysis over large databases look like?

---