Semantic Discovery: Information Retrieval Agents

advertisement
Semantic Web Technologies
• Readings discussion
• Research presentations
• Projects & Papers discussions
Semantic Web ideas aid Discovery
• Machine (system) use of metadata
automatically
• The network is the delivery system &
addressing scheme
- Network properties are metadata too
• Markup languages can express meaning
- Formatting too
- Context matters
• Meaning clusters into taxonomies
- So does use & intent
• Taxonomies can fit into ontologies
- Logical rules define the data & its relationships
Semantic Discovery Challenges
• Is Discovery BIGGER than Search?
• Searching & Understanding
- Interfaces like a reference interview?
• How can we leverage others’ structure?
- Other’s searching?
- Others’ data?
• Does the interface change?
- Make sense of huge amounts of data
- Making sense of data may be more difficult
• Integrating systems logically helps users
understand current information & provides
new information in context
Is it all really about search?
- Discovery of knowledge via taxonomies
- Web service based data searches
- Search by association
• By example
• Facets
- Pattern searches
• OLAP reporting
• Status & monitoring
- Interfaces & agents
• Ease of use
• Automatic use (or help)
- Rule-based queries
• Structured Query Language for structured data?
- Inference
• Personalized search
Berners-Lee on the Semantic Web
• “bring structure to the meaningful
content of Web pages, creating an
environment where software agents
roaming from page to page can readily
carry out sophisticated tasks for users”
(Berners-Lee, 2001)
•
•
•
•
Meaningful?
Content?
Agents?
Tasks?
Searching
Semantic Searching
• Self-describing metadata
- Reliability
• Machine readable, people understandable
• HTML?
- Accuracy
• Personal
• Relevance
- Annotations
• What can information specialists add?
• What other fields?
TAP: Semantic Web Data
• Query Interfaces & Publishing
- Access information from an organized information set
• Semantic Negotiation
- Registries & Trust
• Augmenting with Data
- Denotation
• Ambiguity detection
• Multiple passes
- Extraction
• RDF
- Visualization
• What to show
• “Just knowing that the user is searching form
information about a person… can help avoid several
mistakes.”
Improving search
• What would you do?
Tagging systems for retrieval help
Different kinds of information (images)
Better interfaces for search query formulation
Better interfaces for search results display
Take unstructured information & build taxonomies
or ontologies for organization
- Enable easier comparison of information
• Expose the metadata & make it understandable
- Improve existing search using Web services
-
Agents
• Programs That Use Semantic Web Content
- These programs build upon each other
- Exchanging information between them
• More Automated as Systems and Content
Support Increases
• Digital Signatures
- The Semantics of Trust
- Exchanging proofs
• Service Discovery
• Ontology Exchanges
- Fitting my ontology into yours & sorting out the rules
• Coordination of machines first, then information?
Agents & Trust
• People & machines can understand the data
• Does trust only matter in commerce?
• Extracting structure from multiple searches
-
Can a Semantic Spider help?
• How can you leverage other’s trust when searching?
• How can that be automated with an agent?
• An Ontology for Policies of Trust
-
Permissions
Obligations
Actions
Credentials
Certificates
Semantic Clustering with Facets
• Clustering helps with classifying
• Facets can represent navigable clusters to
help clarify search
- Disambiguation
- Learning about the topic
• Hierarchical Faceted Categories
- Meaningful labels that represent structure &
content
- Navigational control
• Location
• Next steps
• Paths through the information
Download