Semantics on the Web Introduction of Semantic Web

1
Semantics on the Web- Introduction of Semantic
Web
Presented by Zixin Wu Oct 10, 2006
work with Dr. Amit P. Sheth, Dr. John A. Miller,
and other students in LSDIS lab, UGA
2
Outline

• What is Semantic Web?
• The current Web and its problems
• What is semantic? Why do we need it?
• The picture of Semantic Web
• How intelligent Semantic Web could be?
• AI and Semantic Web
• AI in Semantic Web
• Inference
• Applications of Semantic Web
• What are the barriers to realize Semantic Web?

3
The current Web
News from CNN.com
Other types
Audio
Picture from comics.com
XML
Video
PDF
4
The problems of the current Web
Publish
Discover
Consume

• Easy to publish
• Hard to discover (inaccurate result from search
  engines)
• Consumed mainly by humans
• Problems
• Too much information to handle
• Different types of data (heterogeneity)
• Are computers able to help us?

5
Data on the Web

• Current Web (File)
• URL
• Size
• Creator
• Date
• Links to other files

Semantic Web Whats inside the file?
We're moving from a Web of documents to a Web of
data. - Eric Miller
6
Semantic Web A web of resources
5
7
Understand the meaning

• Understand ?
• Actionable knowledge
• Example CNN, News in Chinese
• Structured data
• Machine-processable / readable (unambiguous)
• Is this structured data?
• How about this?
• Have well-defined and relatively simple syntax

8
How to be able to understand?

• Current Web
• Strings (of natural language) in documents
• Only one type of link between documents
• Semantic Web model real world objects and
  phenomenas and their relationships in structured
  data.
• meaning / semantic the relationships between
  concepts.

9
The picture of Semantic Web
Applications
Ontologies (KB)
Meta-data
10
Semantic Web v.s. the current Web
Unstructured data
structured data
Current Web
NLP e.g. QA
Semantic Web
Annotation
Meta-data
11
Semantic Web stack
1
12
Heterogeneity

• On syntax level
• Use uniformed syntax for meta-data
• Such as RDF, OWL
• On semantic level
• Ontology is an agreement
• Ontology mapping / matching

13
AI and Semantic Web

• Two kinds of approaches in AI 2
• Computational Intelligence (iterative development
  or learning)
• Neural networks
• Fuzzy systems
• symbolic AI / logical AI
• Expert systems
• Semantic Web is like a large-scale Expert System
  2
• To solve a well-defined problem by performing
  well-defined operations on existing well-defined
  data

14
Differences between SW and AI

• Not "human-level intelligence
• Distributed
• How the data come into being?
• Heterogeneity
• Inconsistency
• Large-scale
• Scalability
• Open-World Assumption
• Focus on real-world problems instead of toy
  problems

15
AI in Semantic Web 3

• Knowledge representation (description logics)
• Ontology-based reasoning
• Natural language processing
• Machine learning
• Search, Planning
• etc.

16
Inference on a game
Is this a mine?
17
Inference the power of Semantics
Knowledge Base
Facts
new Fact Johns annual income gt 100K
Rules (if X is a member of Richman Club then Xs
annual income gt 100K)
Inference
The really exciting thing happens when others
have their data in a mergeable format and make it
available. When that public information becomes
mergeable, we're in for the next, very pronounced
stage of Web evolution. - Berners-Lee
Fact John is a member of Richman Club
18
Summary of Semantic Web

• Fundamental difference
• A web of data / resources, instead of documents /
  files
• Goals
• Reasoning higher extent of automation
• Integration enhance interoperability
• Features
• Large-scale distributed based on the current web
• Formal representation of resources in structured
  descriptions unambiguous, reasoning
• Uniformed syntax facilitating integration

19
Applications of Semantic Web

• Semantic search more accurate search results
• Swoogle
• Knowledge Management 4
• Ontoknowledge project
• British Telecom Call Center
• Swiss Life
• Cooperation
• Semantic Wiki
• Forum Nokia
• Semantic Blog
• Integrating Business Documents
• Personalization / Personal assistant
• Semantic Bioinformatics / health care 11
• Semantic Web services 78910
• Facilitating data mining

20
Problems of Semantic Web

• Model the world which is infinite?
• Domain specific
• Different schemas
• Ontology mapping / matching
• Reasoning under inconsistency
• Contradictory knowledge bases
• Time-sensitive knowledge
• The cost of developing and maintain of ontologies
  and meta-data 1
• Folksonomies 6
• The Problem of Trust 1 3
• Computational intensive

21
Acknowledgement

• Thanks
• Dr. Amit P. Sheth
• Dr. John A. Miller
• Dr. Prashant Doshi
• Kunal Verma
• Cartic Ramakrishnan
• Some other students in LSDIS

22
References

• 1 A Framework for Web Science
• 2 Semantic Web on Wikipedia.org
• 3 The Semantic Web The Origins of Artificial
  Intelligence Redux
• 4 Semantic Web application areas
• 5 Semantic search
• 6 Folksonomy on Wikipedia
• 7 METEOR-S
• 8 WSMO
• 9 OWL-S
• 10 Semantic Web Services Language
• 11 Glycomics

23
The End

• Thanks and Question?