Outline

Table of Contents

Outline

Logistics

Knowledge Representation

Propositional. Logic vs First Order

Datalog Rules, Programs & Queries

IIIIIS Representation III

Query Containment

Perspective from Logic

Containment Mappings

Computing Containment

Query Planning

Example

Overview of Construction

Inverse Rules

Example

Domain Rules

Priming the Pump

Generating Domain Rules

Resulting Plan

Maximality of Constructed Plan

Outline

Pre-Process Rules

Bottom-up Evaluation

Evaluation

Defining EVAL

Limitations

Outline

Suppose

Efficient & Robust Execution

Source Descriptions

Local Completeness 1 [Etzioni, 94 & Levy, ‘96]

Levy’s Local Completeness

We’d Like to Express

Local Completeness 2

Local Completeness 3

Local Completeness 4

Outline

Plans

Finding Alternatives

Subsumption of Alternatives

Subsumption Proof

Aggressive Execution

Frugal Execution

Summary

Future work:  Decision-theoretic execution

Outline

Inductive Learning of Rules

Types of Learning

Mature Technology

Defining a Learning Problem

Example: Checkers

Example: Handwriting Recognition

Example: Robot Driving

Example: Speech Recognition

Issues

Choosing the Training Experience

Choosing the Target Function

The Ideal Evaluation Function

Choosing Repr. of Target Function

Example: Checkers

Representation

Email: weld@cs.washington.edu

Other information:
CSE 592, Lecture 7

Download presentation source