人工智慧 -- 專家系統

人工智慧

前言

簡介

知識表達

知識學習

理論方法

搜尋優化

邏輯推論

神經網路

機率統計

實務應用

專家系統

自然語言

分群分類

程式語言

Prolog

javascript

程式實作

邏輯推論

爬山算法

基因算法

機率學習

交談程式

數字辨識

訊息

相關網站

參考文獻

最新修改

簡體版

English

Chapter 8: Building a Knowledge Base

Five step of Knowledge Engineering : 
(use circuit design as an example)
1.Decide what to talk about.
Example : 
   Circuit, Terminal, Signal, Gate, Gate-Type (AND, OR ..)

2.Decide on a vocabulary of predicates, functions and constants 
Example : 
   Gate Name         : Gate (X1, X2, .. ..)
   Type Function    : Type(X1), Type(X2).. ..
   Input Predicate : IN(1, X1), IN(2, X2) .. ..

3.Encode general knowledge about the domain.
Example : 
   Connected(t1, t2)=>Signal(t1)=Signal(t2) (同線等電位)
   Signal(t)=On | Signal(t)=Off (不是on就是 Off)
   Type(g)=OR => Signal(Out(1,g))=On <=> Signal(Out(1,g))=Off
   (OR gate的作用)

4.Encode a description of the specific problem instance.
Example :
   設計一個加法器。

5.Pose queries to the inference procedure and get answers.
Example :
   Circuit verification, 用testing case測試是否正確。

本體論(一般型態) General Ontology
1. 分類 Categories : 
By Unary Predicate : Tomato(x) , 
By reification : turning a predicate into an object , 
   Tomatos, is(x, Tomatos)
Disjoint({Animals, Vegetables})
ExhaustiveDecompisition({American, Canadians, Mexicans},NorthAmericans)
Partition({Males, Females}, Animals)

2. 度量 Measures :
Unit function : 
   Length(L1)=Inches(1.5)=Centimeters(3.81)
Order is important : 
   Exercise(e1)&Exercise(e2)&(Difficulty(e1)>Difficult(e2))
   => ExpectedScore(e1)<ExpectedScore(e2)

3. 組合 Composite objects :
PartOf(Bucharest, Romania)
Schema , script : 
PartPartition({head, body, hand, leg}, person)
BunchOf({Apple1, Apple2, Apple3})

4. 時空改變 Time, Space, and Change :
5. 事件與過程 Events and Processes :
Situation calculus的不足:
1. 無法描述連續時間
2. 無法描述 Multiple Agent.

Event calculus : (situation calculus in continuous version)
An Event is just a “chunk” of this universe with both temporal and spatial extent. 

Time Example : SubEvent(BattleOfBritian, WorldWarII)
              SubEvent(WorldWarII, TwentiethCentury)
Place Example: In(NewYork, USA)
Minimization : (描述最小區域或最小時段)
Example: Loaction(x)=l <=> At(x,l)&$At(x,l2)=>In(l,l2)

Process (liquid event categories):
Event(Flying(Johnson, Yesterday))
Throughout(Working(Johnson),TodayLunchHour)
Throughout(Closed(Supermarket1), BunchOf(Sundays))
Throughout(p&q,e) Throughout(p|q, e) Throughout(p xor q,e)
----------------            ---   ---
----------------    --------------         --    ---
Interval(i)=>(Moment(i)<=>Duration(I)=0)
Interval(i)=>Duration(i)=(Time(End(i))-Time(Start(i)))

Meet(i,j)<=> Time(End(I))=Time(Start(j))
Before(i,j)<=> Time(End(I))<Time(End(j))
After(j,i)<=> Before(i,j)
During(i,j)<=> Time(Start(j))<=Time(Start(I))&Time(End(I)<=Time(End(j)))
Overlap(i,j)<= > #k During(k,i)&During(k,j)

Meet(i,j)     i: -----------
                 j:              --------------
Before(i,j)     i: --------    
               j:              --------------

During(i,j)     i:    --------
                 j: --------------

Overlap(i,j)    i: --------------
                   j:    --------------

6. 實體 Physical Objects :
Typical典型 : Typical(C) is subset of C : 
  若無特別指定(例外),則x in C => x in Typical(C)
Fluents Object 會隨時間改變的實體(temporal substance): 
example : President(USA)
individuation 不可分割的實體 : 
example : People

7. 物質 Substances :
可分割的實體stuff (spatial substance) (example : Butter) 
    Butter&PartOf(y,x)=>Is(y,Butter)

8. 精神、信仰 Mental Objects and Beliefs
Believes(Agent,x) : x is referential opaque
example : 
  Superman=Clark ╞ (Believes(Lois,  Flies(Superman))<=>
                       (Believes(Louis,Flies(Clarkis)))
  is not right.

Referential transparency : 
The property of begin able to freely substitute a term for an equal
 term is called referential transparency.

Syntactic theory : Flies(clark) => [F,l,i,e,s,(,C,l,a,r,k,)]
Semantic : 
  Denote(“Clark”)=ManOfSteel&Denote(“Superman”)=ManOfSteel
  Name(ManOfSteel)=”K11”
Proof Theory :
  LogicalAgent(a)&Believes(a,p)&Believes(a,”p=>q”)=>Believes(a,q)

Facebook

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License