Modeling Reasoning with Making it uncertainty uncertainty - - PowerPoint PPT Presentation

Handling Incomplete, Uncertain, Inconsistent Data

• Important subject: such data is everywhere
• Key research challenges
• Tools/techniques we need
• Neighbors
• Some more general thoughts

Modeling uncertainty

• Types of uncertainty
• what is the truth?
• How to store/

represent it?

• Ranking uncertain

answers

• Connection to
• prob. programming

Reasoning with uncertainty

• Learning queries on

uncertain data

• Inferences with

incomplete data

• Generalizability of

query results

• Integrating different

types of uncertainty

Making it practical

• Approximate

answers

• Query optimization

with incomplete info

• Inconsistency/CQA:

making it work for data cleaning

• Practical query

evaluation on prob. data

Tools/techniques: what is needed

• Benchmarks: we don’t really have any
• but who will maintain them?
• Approximate solutions
• there is life beyond CQs!
• Downloadable software tools

Neighborhs

• SIGMOD!!!
• Uncertainty in AI: UAI/SUM/STARAI...
• Traditional AI: KR, DL, AAAI, IJCAI...

More general points

• SIGMOD does not see us (well)
• PR/visibility is important!
• ACTIONS
• PODS Day 1: it’s happening now, PODS EXEC - act!!!
• Research Highlights: CACM, now SIGMOD Record
• More involvement in SIGMOD events