›
›

Bài giảng Trí tuệ nhân tạo - Chapter 4: Knowledge Representation

Download
Knowledge Representation  
Chapter 4  
What is KR?  
R. Davis, H. Schrobe, P. Szolovits (1993):  
1. A surrogate  
2. A set of ontological commitments  
3. A fragmentary theory of intelligent reasoning  
4. A medium for efficient computation  
5. A medium of human expressions  
2
Representation and Mapping  
• Facts: things we want to represent.  
• Representations of facts: things we can manipulate.  
3
Representation and Mapping  
reasoning  
programs  
Internal  
Facts  
Representations  
English  
understanding  
English  
generation  
English  
Representations  
4
Representation and Mapping  
desired real reasoning  
Initial  
facts  
Final  
facts  
forward  
representation  
mapping  
backward  
representation  
mapping  
Internal  
representations  
of initial facts  
Internal  
representations  
of final facts  
operation  
of program  
5
Representation and Mapping  
• Spot is a dog  
• Every dog has a tail  
Spot has a tail  
6
Representation and Mapping  
• Spot is a dog  
dog(Spot)  
• Every dog has a tail  
∀
x: dog(x)  
→
hastail(x)  
hastail(Spot)  
Spot has a tail  
7
Representation and Mapping  
• Fact-representation mapping is not one-to-one.  
• Good representation can make a reasoning program  
trivial.  
8
Representation and Mapping  
The Multilated Checkerboard Problem  
“Consider a normal checker board from which two  
squares, in opposite corners, have been removed.  
The task is to cover all the remaining squares exactly  
with donimoes, each of which covers two squares. No  
overlapping, either of dominoes on top of each other or  
of dominoes over the boundary of the multilated board  
are allowed.  
Can this task be done?”  
9
Representation and Mapping  
No. black squares  
= 30  
No. white square  
= 32  
10  
Representation and Mapping  
Good representation:  
• Representational adequacy  
• Inferential adequacy  
• Inferential efficiency  
• Acquisitional efficiency  
11  
Approaches to KR  
Simple relational knowledge:  
• Provides very weak inferential capabilities.  
• May serve as the input to powerful inference engines.  
12  
Approaches to KR  
Inheritable knowledge:  
• Objects are organized into classes and classes are  
organized in a generalization hierarchy.  
• Inheritance is a powerful form of inference, but not  
adequate.  
13  
Approaches to KR  
Inferential knowledge:  
• Facts represented in a logical form, which facilitates  
reasoning.  
• An inference engine is required.  
14  
Approaches to KR  
Procedural knowledge:  
• Representation of “how to make it” rather than “what  
it is”.  
• May have inferential efficiency, but no inferential  
adequacy and acquisitional efficiency.  
15  
Approaches to KR  
Choosing the Granularity:  
• High-level facts may not be adequate for inference.  
• Low-level primitives may require a lot of storage.  
16  
Homework  
Reading  
R. Davis, H. Schrobe, P. Szolovits (1993): “What is a knowledge  
representation?”  
17  
pdf 17 trang myanh 20140 Free
Download
Bạn đang xem tài liệu "Bài giảng Trí tuệ nhân tạo - Chapter 4: Knowledge Representation", để tải tài liệu gốc về máy hãy click vào nút Download ở trên

File đính kèm:

  • pdfbai_giang_tri_tue_nhan_tao_chapter_4_knowledge_representatio.pdf