ai-lect9 - Knowledge Representation Knowledge engineering:...

Info iconThis preview shows pages 1–11. Sign up to view the full content.

View Full Document Right Arrow Icon
1 Knowledge Representation Knowledge engineering: principles and pitfalls Ontologies Examples
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
2 Knowledge Engineer Populates KB with facts and relations Must study and understand domain to pick important objects and  relationships Main steps: Decide what to talk about Decide on vocabulary of predicates, functions & constants Encode general knowledge about domain Encode description of specific problem instance Pose queries to inference procedure and get answers
Background image of page 2
3 Knowledge engineering vs. programming Knowledge Engineering Programming 1. Choosing a logic Choosing programming language 2. Building knowledge base Writing program 3. Implementing proof theory Choosing/writing compiler 4. Inferring new facts Running program Why knowledge engineering rather than programming? Less work:  just specify objects and relationships known to be true, but  leave it to the inference engine to figure out how to solve a problem  using the known facts.
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
4 Properties of good knowledge bases Expressive Concise Unambiguous Context-insensitive Effective Clear Correct Trade-offs:  e.g., sacrifice some correctness if it enhances brevity.
Background image of page 4
5 Efficiency Ideally:  Not the knowledge engineer’s problem The inference procedure should obtain same answers no matter  how knowledge is implemented. In practice: - use automated optimization - knowledge engineer should have some understanding of how inference is done
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
6 Pitfall: design KB for human readers KB should be designed primarily for inference procedure! e.g., VeryLongName  predicates: BearOfVerySmallBrain(Pooh) does not allow inference procedure to  infer that Pooh is a bear, an animal, or that he has a very small  brain, … Rather, use: Bear(Pooh) 2200 b, Bear(b)   Animal(b) 2200 a, Animal(a)  PhysicalThing(a) [See AIMA pp. 220-221 for full example]
Background image of page 6
7 Debugging In principle, easier than debugging a program, because we can look at each logic sentence in isolation and tell  whether it is correct. Example: 2200 x, Animal(x)    5  b, BrainOf(x) = b means  “there is some object that is the value of the BrainOf function  applied to an animal”   and can be corrected to mean  “every animal has a brain”   without looking at other sentences.
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
8 Ontology Collection of concepts and inter-relationships Widely used in the database community to “translate” queries and  concepts from one database to another, so that multiple databases  can be used conjointly (database federation)
Background image of page 8
9 Ontology Example Khan & McLeod, 2000
Background image of page 9

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
10 Towards a general ontology Develop good representations for: - categories - measures - composite objects - time, space and change - events and processes - physical objects - substances - mental objects and beliefs -
Background image of page 10
Image of page 11
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 01/20/2011 for the course CS 6810 taught by Professor Hecker during the Spring '10 term at CSU East Bay.

Page1 / 54

ai-lect9 - Knowledge Representation Knowledge engineering:...

This preview shows document pages 1 - 11. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online