150 likes | 171 Views
This course provides an overview of knowledge representation and reasoning, discussing the historical background, different levels of knowledge, early AI enthusiasms, and the future of KR. Students will learn how to represent knowledge precisely using state-of-the-art representation schemes and program in Prolog.
E N D
Overview • Contact Information • Grading policy • Syllabus • How to fail this class • What is knowledge?
General Information • Instructor: Lee McCauley 374 Dunn Hall 678-2486 mccauley@memphis.edu Office Hours: Tue. & Thur. 1:00 – 2:30
Evaluation • Class Participation = 20% • Model Presentations = 20% • Project Write-up = 20% • Project Code/Demo = 20% • Homeworks = 20%
Syllabus • Refer to your handouts
How to fail this class • Don’t show up • Don’t take the project seriously • Don’t do the homework • Come ill-prepared for presentations
What is knowledge? Discussion
Historical Background • Socrates began the art of rhetoric in the fifth century BC – and died for corrupting the minds of Athenian youth • Plato, his student, created the field of epistemology – the study of the nature of knowledge • Aristotle, Plato’s student, (who did not want to die) shifted the emphasis from the nature of knowledge to the representation of knowledge
Tentative Definition • Justified belief that increases an entities capacity for effective action (Nonaka 1994, Huber 1991)
Data, Information, and Knowledge • Data • Raw material/sensation • Information • Categorized data • Data with meaning that may change knowledge • Knowledge • Actionable information • What to do with the information • Information that can be reasoned to be either true or false
Early AI enthusiasms • Logic and theorem proving eagerly adopted • Computational issues forced consideration of how to package up knowledge, control inference • Frame languages • Special-purpose KR languages • Formalists versus Hackers
Form minus content • Movement in 1980s: KR = Formal KR • Reaction to lack of clear semantics • Identification of formality with precision • Focus on general logical schemes, not specific domains • Consequences • Lots of technical progress • Common perception of sterility in many areas, e.g. nonmonotonic logics • Most exciting KR work didn’t appear in KR community, e.g., qualitative physics, CYC project, ...
The Representation Resurgence • Representation Lite hits too many walls • Web search engines adding more semantics along with statistical techniques • Dramatic success stories in narrow areas • Scheduling: Desert Shield, Detecting money laundering, Detecting stolen credit cards… • Steady scientific progress in AI • KR now embracing content again • Moore’s law is making it all practical
The Future of KR • Ideas, technologies, and tools now coming together • Clear perception arising of need for common sense knowledge bases • Keeping up with the Web – NLP rises again! • See Semantic Web, DAML projects • Software that you treat as a collaborator • Knowledge management • The infrastructure is being created today • Those who understand KR will shape what happens
What this course is about • You will learn how to represent knowledge very precisely • So precisely that computer programs can use it • You will learn state of the art representation schemes for core kinds of knowledge • Space, time, quantity, events, causality, common sense… • You will learn how to program in a powerful KR language - Prolog