Knowledge Representation

Slides:



Advertisements
Similar presentations
Expert System Seyed Hashem Davarpanah
Advertisements

Artificial Intelligence
1 Knowledge Representation Introduction KR and Logic.
ARTIFICIAL INTELLIGENCE [INTELLIGENT AGENTS PARADIGM] Professor Janis Grundspenkis Riga Technical University Faculty of Computer Science and Information.
Knowledge Representation
Intelligent systems Lection 7 Frames, selection of knowledge representation, its combinations.
Chapter 4 Knowledge Representation Artificial Intelligence ดร. วิภาดา เวทย์ประสิทธิ์ ภาควิชาวิทยาการคอมพิวเตอร์ คณะ วิทยาศาสตร์ มหาวิทยาลัยสงขลานครินทร์
Knowledge Representation. Essential to artificial intelligence are methods of representing knowledge. A number of methods have been developed, including:
CPSC 322 Introduction to Artificial Intelligence November 5, 2004.
1 Knowledge Representation We’ve discussed generic search techniques. Usually we start out with a generic technique and enhance it to take advantage of.
Semantic Nets, Frames, World Representation. Knowledge Representation as a medium for human expression An intelligent system must have KRs that can be.
For Friday Finish chapter 10 No homework (get started on program 2)
14th September 2006 Dr Bogdan L. Vrusias
Knowledge Representation
Knowledge Engineering
 Contrary to the beliefs of early workers in AI, experience has shown that Intelligent Systems cannot achieve anything useful unless they contain a large.
Knowledge Representation
Knowledge Representation and Objects
Objects Objects are at the heart of the Object Oriented Paradigm What is an object?
Knowledge representation methods جلسه سوم. KR is AI bottleneck The most important ingredient in any expert system is knowledge. The power of expert systems.
ARTIFICIAL INTELLIGENCE [INTELLIGENT AGENTS PARADIGM] Professor Janis Grundspenkis Riga Technical University Faculty of Computer Science and Information.
Knowledge Representation
Lectures 5,6 MACHINE LEARNING EXPERT SYSTEMS. Contents Machine learning Knowledge representation Expert systems.
INF 384 C, Spring 2009 Ontologies Knowledge representation to support computer reasoning.
Artificial Intelligence 4. Knowledge Representation Course V231 Department of Computing Imperial College, London © Simon Colton.
For Friday Exam 1. For Monday No reading Take home portion of exam due.
Alternative representations: Semantic networks
Knowledge Representation CPTR 314. The need of a Good Representation  The representation that is used to represent a problem is very important  The.
For Wednesday Read chapter 13 Homework: –Chapter 10, exercise 5 (part 1 only, don’t redo) Progress for program 2 due.
Early Work Masterman: 100 primitive concepts, 15,000 concepts Wilks: Natural Language system using semantic networks Shapiro: Propositional calculus based.
Artificial Intelligence LECTURE 2 ARTIFICIAL INTELLIGENCE LECTURES BY ENGR. QAZI ZIA 1.
Lectures on Artificial Intelligence – CS435 Conceptual Graphs
ARTIFICIAL INTELLIGENCE [INTELLIGENT AGENTS PARADIGM] Professor Janis Grundspenkis Riga Technical University Faculty of Computer Science and Information.
Structured Knowledge Chapter 7. 2 Logic Notations Does logic represent well knowledge in structures?
 Dr. Syed Noman Hasany 1.  Review of known methodologies  Analysis of software requirements  Real-time software  Software cost, quality, testing.
Semantic web course – Computer Engineering Department – Sharif Univ. of Technology – Fall Knowledge Representation Semantic Web - Fall 2005 Computer.
KNOWLEDGE REPRESENTATION 6 6.0Issues in Knowledge Representation 6.1A Brief History of AI Representational Systems 6.2Conceptual Graphs: A Network Language.
Semantic Nets, Frames, World Representation CS – W February, 2004.
Intelligent Control Methods Lecture 7: Knowledge representation Slovak University of Technology Faculty of Material Science and Technology in Trnava.
Knowledge Representation
Some Thoughts to Consider 8 How difficult is it to get a group of people, or a group of companies, or a group of nations to agree on a particular ontology?
1 CS 385 Fall 2006 Chapter 7 Knowledge Representation 7.1.1, 7.1.5, 7.2.
For Wednesday Read chapter 13 No homework. Program 2 Any questions?
Finite State Machines (FSM) OR Finite State Automation (FSA) - are models of the behaviors of a system or a complex object, with a limited number of defined.
Knowledge Representation Fall 2013 COMP3710 Artificial Intelligence Computing Science Thompson Rivers University.
Lecture 5 Frames. Associative networks, rules or logic do not provide the ability to group facts into associated clusters or to associate relevant procedural.
1 Frame Theory A vague paradigm - to organize knowledge in high-level structures “A Framework for Representing Knowledge” - Minsky, 1974 Knowledge is encoded.
Artificial Intelligence Knowledge Representation.
Knowledge Engineering. Sources of Knowledge - Books - Journals - Manuals - Reports - Films - Databases - Pictures - Audio and Video Tapes - Flow Diagram.
Definition and Technologies Knowledge Representation.
16 April 2011 Alan, Edison, etc, Saturday.. Knowledge, Planning and Robotics 1.Knowledge 2.Types of knowledge 3.Representation of knowledge 4.Planning.
Artificial Intelligence Logical Agents Chapter 7.
Knowledge Representation Part I Ontology Jan Pettersen Nytun Knowledge Representation Part I, JPN, UiA1.
Artificial Intelligence 4. Knowledge Representation
Knowledge Representation
Knowledge Representation Techniques
Artificial Intelligence
Knowledge Representation
Knowledge Representation
Knowledge Representation
Artificial Intelligence (CS 370D)
Knowledge Representation
KNOWLEDGE REPRESENTATION
Weak Slot-and-Filler Structures
Knowledge Representation
Semantic Nets and Frames
Structured Knowledge Representation
Subject : Artificial Intelligence
Habib Ullah qamar Mscs(se)
Presentation transcript:

Knowledge Representation

CONTENTS What is knowledge? How to represent knowledge? Definition of KR Characteristics of KR Schemes Types of KR Schemes: Logic Procedural Network Structured Types of KR Methods: Semantic Network Frame Script Conceptual Graph References

What is knowledge? Knowledge is information/fact about some domain, subject area or about how to do something. Knowledge can take many form. Some examples: Eve is a female and Adam is a male. All females with children are mothers. Mothers are females, fathers are males. 2 x 5 = 10 and 5 x 2 = 10. All students in UOFG are brainy. Jerry is a mouse and Tom is a cat. No mouse is bigger than a cat.

How to represent knowledge? Can we use natural languages to represent knowledge? Advantage: Natural languages is expressive enough. Disadvantages: Too ambiguous for automated reasoning. Semantic ambiguities: Time flies like an arrow. Pretty little girls’ school. English teacher. Syntactic ambiguities: Bark in dog’s bark and tree bark. Right in turn right and you’re right. Bank in river bank and bank the money. Time moves quickly just like an arrow does? Measure the speed of a flying insects like you would measuring that of an arrow? Measure the speed of flying insects like an arrow would? Measure the speed of flying insects that are like arrows? A type of flying insects i.e. “time-flies” that enjoy arrows? Does the school look little? Do the girls look little? Does the school look pretty? Do the girls look pretty? January 2008 / Norshuhani Zamin A teacher who teaches English? A teacher from England?

How to represent knowledge? Can we use databases to represent knowledge? Yes, databases have been the most common option for software application to store and manipulate virtually any kind of data. Advantage: Well suited to efficiently representing and processing large amount of data. Disadvantages: Only simple problem domains can be accommodated. Entities, relationship between entities can be represented but not much more. Very simple reasoning i.e. simple lookup using SQL statements. Very structured type records and hard to manipulate.

How to represent knowledge? A record structure VS graph representation

Definition of KR KR is the study on how information related to cognitive sciences can be appropriately encoded and utilized in computational models of cognition. KR closely related with mental representation, deductive reasoning, philosophy of language and logic. Also known as Knowledge Engineering. KR has created a collection of formalized representational schemes and methods applied in most of the previous and current AI applications. Creating appropriate knowledge representation schemes and methods for AI is not easy.

Characteristics of KR Schemes KR scheme is the framework used to represent some categories of knowledge. There are four characteristics of KR schemes: Representational Adequacy Ability to represent all necessary knowledge in a particular domain. Inferential Adequacy Ability to infer i.e. generate new knowledge from old knowledge. The inferences made should be: Sound: The new knowledge does follow from the old knowledge. Complete: Making the right and logical inference. Inferential Efficiency Ability to combine new knowledge which can be used by inference mechanism in determining the best direction based on the current knowledge structure. Acquisition Efficiency Ability to add new information into knowledge base (by user or program)

Types of KR Schemes KR schemes can be classified into four categories: Logic Use formal logic to represent knowledge. Examples: propositional and predicate calculus. Procedural Represent knowledge with a set of sequential instructions to solve problems Examples: flow chart, pseudocode, production rule, script. Network Represent knowledge as graph in which the nodes represent object/concept in the problem domain and arcs represent relations/associations between them. An attempt to incorporate human memorizing ability into AI. Examples: semantic network, state space, mind map (assignment). Structured Extend networks by allowing each node to be a complex data structure consisting of property types and values. Example: frame.

Types of KR Methods: Semantic Network Introduced by Quillian in 1968 as a model of human memory. A knowledge representation in the form of graph. The nodes of the graph correspond to facts. The link of the arcs represent the relationships or associations between the facts. Both nodes and arcs are labeled. Semantic network supports inheritance.

Types of KR Methods: Semantic Network Example: Some possible inferences: Bill is a cat that has 4 legs. Bill is a mammal Opus is a type of bird. Opus is a penguin. Opus cannot fly.

Types of KR Methods: Semantic Network 4 More examples: 1 3 January 2008 / Norshuhani Zamin 2

Types of KR Methods: Semantic Network Advantages: Easy to visualize. Relationship can be arbitrary defined by knowledge engineer. Related knowledge easily clustered. Efficient in space requirement. Object defined only once (inheritance concept). Disadvantages: Unwanted inheritance may cause problems. Inappropriately placed facts can cause problems. No standard about node and arc values.

Types of KR Methods: Frame Introduced by Marvin Minsky in 1974. It extends semantic network to provide a more structured way of representing a knowledge base. It stores properties, values, methods and relevant information of object. Frame supports class hierarchies applied in object oriented concept. Each frame has: A name A slot which stores information like specific value, default value, inherited value, a pointer to another frame (superclass or subclass)

Types of KR Methods: Frame Example 1:

Types of KR Methods: Frame Animals Alive T Example 2: Flies F isa Birds Mammals Legs 2 Legs 4 Flies T isa Penguins Cats Bats Flies F Legs 2 Flies T instance Opus Bill Pat Name Opus Name Bill Name Pat Friend Friend

Types of KR Methods: Frame Advantages: Very flexible for many type of applications. Similar to human knowledge organization. Suitable for causal knowledge. Easy to include default information and detect missing values. Easier to understand than logic or rules. Disadvantages: No standard for slots filler values. More on general representation than a specific representation. For example, frame of a class room for primary and tertiary education should be different. No associated reasoning or inference mechanism.

Types of KR Methods: Script A structured representation describing a stereotyped sequence of events related to a particular context introduce by Roger Schank in 1977. They are used in natural language understanding to understand specific situation. Components of script: Entry condition conditions that must be true for the scripts to be valid. Results facts that become true once the script is completed. Props secondary things that support the script. Roles actions that individual participants perform. Scenes there are many scenes in a script. Each one represents a specific temporal aspect of the script.

Types of KR Methods: Script For example, a script for the process.. “Going to a restaurant to have a meal”

Types of KR Methods: Conceptual Graph Conceptual Graph (CG) is finite, connected bipartite graph or bigraph. A bipartite graph is a graph in which the set of nodes can be partitioned in two disjoint sets. There 2 types of node in the CG: Concept: the knowledge / fact / action Relation: the type of relationship between 2 concepts. Rule in CG: there are NO arcs between a concept and another concept, and no arcs between a relation and another relation. All arcs either go from a concept to a relation or from a relation to a concept. CONCEPT A RELATION CONCEPT B CONCEPT A CONCEPT B RELATION A RELATION B Valid CG Invalid CG

Types of KR Methods: Conceptual Graph Examples: A dog is brown A cat is on the mat dog color brown cat on mat A monkey scratch its ear with a paw part of monkey scratch agent object ear instrument paw part of

Types of KR Methods: Conceptual Graph General Concept VS Specific Concept A cat is grey General concept CG is referring to a particular but unknown instance. Cat: X color grey A cat named Abu is grey Cat: Abu color grey Cat color grey Specific concept CG is referring to a particular and known instance. name “Abu”

Types of KR Methods: Conceptual Graph Propositional Concept: All people like pizza Person:  agent like object pizza Ali believes that all people likes pizza Person: Ali agent believes object proposition: Person:  agent like object pizza There are no pink dogs proposition: negation dog color pink

Types of KR Methods: Conceptual Graph Joining Concept: Doraemon eats dorayaki greedily cat:Doraemon person agent eat object pie dorayaki pie pie pie pie pie cat:Doraemon agent eat manner pie greedy pie pie pie pie pie object pie pie dorayaki pie pie pie pie cat:Doraemon agent eat manner greedy

References Books: 1. Artificial Intelligence: Structures and Strategies for Complex Problem Solving Author: George F Luger Publisher: Addison Wesley 2. Textbook – Chapter 10 Websites: http://www.cs.umbc.edu/771/papers/hayesp.html http://www.cs.cf.ac.uk/Dave/AI2/AI_notes.html http://www.jfsowa.com/cg/cgexampw.htm Good thesis discussed how to represent CG in programming: http://staff.science.uva.nl/~gilad/pubs/MoL-2003-04.text.pdf