music & knowledge sharing
Semantic Web overloaded
- Characterize Today's Semantic Web?
- What do you think is 'Semantic Web' ?
- Different people use 'Semantic Web' differently
- quite vague and overloaded - SF like dream
- We cannot "do" Semantic Web
- Like "I did the Internet yesterday" instead of "I browsed the WWW yesterday"
- Huge potential, but not bootstrap yet
- some innovators play only in their fields
- not business first, but users first
- See WWW, bottom up
Some facets of SemWeb
- Advanced Search
- with metadata
- first step to make SW mainstream
- Data reuse and interoperability
- with URI vocabulary and ontology
- everybody can join
- Intelligent Agent
- with Ontology and Rules: SWS
- Web of Trust
Be easy, fun and useful
- Key is users, not researchers
- It should be fun and/or useful
- weblog is fun
- social net is fun and useful
- social bookmark is useful
- It should be easy to use
- Many SW apps are still look difficult
- no unix make/install, no java install
- web based apps get popular
Advanced Search
- Most straightforward way to SW be mainstream
- because it is useful
- with metadata, more precise and comprehensive result
- Chicken and egg problem
- metadata first, or metadata search engine first
- providing metadata is difficult and low motivation
- Data not necessarily be RDF based
- RDF triple is very powerful, flexible
- but RDF/XML looks difficult and redundant
- Simple mapping rule will work
- btw. coloq XML/RDF, XHTML/RDF etc.
Folksonomy
- Folksonomy is so hot because
- it's easy and intuitive
- tags are practical for some discoveries
- people want method for advanced search
- people enjoy that they can contribute
- Can floksonomy be an answer ?
- controlled vocab. vs. free term
- universal vs domain specific
- very important step toward SW
- can work with formal, controlled vocab.
- machine can help like clustering
Possible Viral Service
- Folksonomy based search
- some RSS/weblog search
- tag by others
- free keywords in dc:subject
- extraction via GRDDL
- Metadata extraction service
- machine generated metadata
- Metadata based search
… more possibilities
- More automated metadata generation
- weblog plugins
- standard thesauri and assist tools
- Automated metadata services
- Desktop integration
- MacOS Tiger and Longhorn provide metadata for files
- possibly add basic metadata on upload
Reuse and Interoperability
- Politically Correct
- Combination of multiple vocabularies
- Query over multiple data sources
- but how can users find them easy, fun and useful?
- Need attractive apps/services like:
- RSS + FOAF + GEO + iCal
- find a concert info from a calendar, and find restaurant near the venue using geo info
- People can join
- distributed, bottom up + aggregate
- anyone can easily enter
URI based vocabulary
- How to integrate non-URI-based vocabularies
- microcontents
- XML based: SVG etc ?
- Needs simple rule to construct RDF URI
- NS URIs not always end with / or #
- cannot distinguish local name from URI based term
- some standard mapping rule is desired
- maybe simply add # after NS URI
- can integrate these vocab into the Semantic Web
Possible approaches
- Integrate FOAF and SNS + microformat
- Integrated event calendar
- Upcoming.com provides icalendar + folksonomy
- Google map annotation
- Moblog + GPS
Intelligent Agent ?
- Not sure
- no standard rule yet
- much harder to provide rules than just metadata
Who'll get the most from SW?
- Must be users
- users first
- will benefit industries as well
- Enterprise can gain from SW like tech
- could be an IP based AI
- SW is in open world, enterprises are in closed world
- SW must consider trust
