Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
knowledge_representation [2006/06/10 18:05]
Charles Chen
knowledge_representation [2016/10/22 20:12] (current)
Line 1: Line 1:
 +== Knowledge representation [知識表徵]==
 +**Definitions**:​
  
 +//Knowledge representation is a umbrella term for the methods by which information is stored in the mind for later use.// - by Arthur B. Markman, 2005
 +
 +//Knowledge representation is a subarea of artificaial intelligence concerned with understanding,​ designing, and implementing ways of representing information in computers so that programs can use this information to: derive information that is implied by it; to converse with people in natureal languages; to plan future activities; and solve problemsin areas that normally require human expertise.//​ - by Stuart C. Shapiro, 2005
 +
 +**book**: //Knowledge representation//​ by Arthur B. Markman, 1999 http://​books.google.com/​books?​id=Ytys7EgskXsC ​
 +
 +「表徵」一詞在不同的學科中,有不同的意涵。在1950s後期,心理學開始以電腦的觀點研究心智。這種觀點假設人類的心智就像電腦一樣,以表徵完成演算工作,以此觀點來說明人類心智處理資訊的工作。在心理學中,廣義的表徵(representation)可以用四個部分來定義(Markman,​ A., 1999):
 +  - a represented world / 被表徵的世界
 +  - a representing world / 表徵的世界
 +  - a set of representing rules / 表徵的規則
 +  - a processe that uses the representation / 使用表徵資訊的程序
 +
 +
 +
 +
 +== Introduction ==
 +
 +=== 知識表徵(系統)(人工智能) ===
 +知識表徵是一個人工智能研究的次領域,探討理解、設計、與以計算機實作表徵資訊的方式。因而得以:
 +  * 導出隱含其中的資訊;
 +  * 與人類以自然語言交談;
 +  * 計劃未來行動;
 +  * 解決通常需要人類專門知識處理的問題
 +
 +導出(derviing)資訊中隱含的資訊以推論(reasoning)的形式表徵。因此這個領域的研究又被稱為「知識表徵與推論」。
 +
 +許多哲學家認為知識(knowledge)即被證明(justified)為真的信念(true belief)。但是在人工智慧的知識表徵研究中,對知識的定義並不侷限在如此嚴謹的哲學定義中。
 +
 +==== 歷史 ====
 +知識表徵研究開始於1958年左右,目標在於使電腦能夠處理人類所能進行的常識判斷。1960s-1970s的知識表徵研究,認為人們都是以自然語言處理知識表徵與資訊的使用。1970s-1980s的研究關注於人類專業作業的特定領域知識。1980s-1990s,研究轉為關注於日常生活知識中的特定次領域細節,如時空表徵(spatial representation)、一般目的的知識架構(ontology;​ Cyc Project)。
 +
 +== 知識表徵的種類 ==
 +  - spatial representation 時空表徵
 +  - featural representation 特徵表徵
 +  - semantic network 語義網路
 +  - structured representation 結構表徵 ​
 +
 +== predicate calculus ==
 +predicate calculus - 謂詞演算
 +
 +在1800s-1900s初期,發展出許多希望能將人類的推論邏輯加以計算的形式系統(formal system),或可統稱唯一種形式邏輯系統。對人工智能的研究者而言,這是知識表徵系統的濫觴。這時期所發展的系統由三種基本要素組成:
 +  - syntax - 語法: 規範基本符號與形式表徵語法規則
 +  - semantics - 語義: 規範基本符號的意義,與由各部分決定意義表徵的規則
 +  - proof theory - 證明論: 規範推論的規則,指定哪個其他表徵可以被加入到公理(axioms)的一個起始集合「證明」(an initial collection called "​proof"​)。
 +
 +meaning
 +
 +在邏輯系統中有兩種meaning:​
 +  - 句子所聲稱的
 +  - 真值(true value)
 +
 +propositional logic 與 predicate logic (first-order logic) 在邏輯語法上的不同
 +
 +== procedural representation ==
 +In the mid-1970s, knowledge representation researchers were embroiled in what was called the '​declarative/​procedural controversy'​. ​ \\ know-what/​know-how \\ 
 +
 +
 +== production system ==
 +Production systems are a subclass of knowledge representation and reasoning systems. The knowledge base of a production system is divided into two parts, a working memory and a rule memory. The working memory consists of a set of symbol structures not containing varialbes. The rule memory consists of a set of pattern-action rules.
 +
 +== references ==
 +
 +  * //Stuart C. Shapiro// -  http://​www.cse.buffalo.edu/​~shapiro/​
 +  * Shapiro, S. C. (2005). Knowledge reresentation. In L. Nadel (Ed.), Encyclopedia of cognitive science (pp. 671-680). NJ: John Wiley.
 +  * Markman, A. B. (1998). //Knowledge representation//​. Mahwah, NJ: Lawrence Erlbaum Associates Inc.
 +  * Markman, A. B. (2005). Knowledge reresentation,​ psychology of. In L. Nadel (Ed.), Encyclopedia of cognitive science (pp. 680-687). NJ: John Wiley.