Transcript Phil Bernstein - Microsoft Research

Lowell Gong Show
Phil Bernstein
May 2, 2003
Phil Bernstein
1
Drivers of Research Topics
1.
Exploit Improved System Technology

networks (dist’d DB), WWW (data integration),
disks (VLDBs), sensors (streaming queries)
2. New Ideas for Classical DB Technologies

access methods, query optimiz’n, auto-admin, metadata
3. Grand Challenges

Info Utility (Asilomar’98), Jim Gray’s list, CRA GC Conf
no medical errors (Lomet), no auto fatalities (Lampson)
4. Combine DB & Non-DB technology

natural language, rules, data mining, XML
5. Applications

May 2, 2003
engineering, multimedia, bio-medical informatics
Phil Bernstein
2
Knowledge-Based Systems that Work
► Move
from data mgmt to knowledge mgmt
 A useful way to soak up desktop cycles
 A big functionality improvement over DBMSs
► More
logical inferencing
 Theorem proving, fuzzy inferencing, model checking
 Integrated with query processing and IR
► Richer
knowledge representation
 And much larger knowledge bases
► Engineer
the combination of AI + DB
 Robust systems with predictable capability
► Applications
– scientific investigation,
engineering design, medical diagnosis
May 2, 2003
Phil Bernstein
3
Experiment Management Systems
► Data
mgmt for experimental investigation
 Rapid evolution of system under test, schemas,
analysis scripts, workflow
 Comprehensive versioning (types, schemas, data)
 Heavy use of rich data types
 Heavy use of data mining
 Integration with portal management
► Applications
 Science – especially bio-medical
 Engineering – benchmarking, process engineering
May 2, 2003
Phil Bernstein
4