Data.gov: Web, Data Web, Social Data Web [email_address] 7/22/2010 #health2stat
- From a Web of Linked Documents,
- Bear with me for a quick overview -
- There’s only one type of link in HTML
- SourcePage.htm contains markup;
- <a href='target.html'> hyperlink </a>
- <myTag> typed data </myTag>
- Good, but more importantly...
- The Architecture of the Web
- (one of the most important and successful inventions of the 20 th Century!)
- (Info) Resource Representations (URI’s vs URL’s), Uniform Interface (CRUD via HTTP methods), server Resource and client App state (via links)...
- The Data Web (aka Web of Data) is an innovation that extends the existing Web of Documents
- Beginning with the Resource Description Framework ( RDF )
- Which is a Web based data modeling language
- RDF gives us custom link types :
- thing <relationship> thing
- subject <predicate> object
- HTTP GET ’dereferences’ RDF triples in multiple serialization (.htm, .rdf, .json) formats
- has a URL representation that returns
- <http://my.org/predicate> <http://your.com/Object>
- Grammar needs vocabularies ...
- Vocabularies can be metadata :
- Objective quantifies Goal
- So http://standards.org/vocab/Objective returns;
- <quantifies> <http://standards.org/vocab/Goal>
- Instance data (re)uses vocabularies:
- Meaningful Use 'is a' (type of) Objective
- <type> <http://standards.org/Objective>
- EHR Exchange 'is a' (type of) Goal
- <type> <http://standards.org/Goal>
- Now we’ve introduced to the what , but -
- WHY should we do this Linked (Open Gov) Data stuff?
- We can (automatically) infer things like;
- Meaningful Use quantifies EHR Exchange
- We can traverse (via apps and browsers) the data graph with no apriori domain knowledge …
- We/others just make links to correlate disparately owned/managed/published data across distinct (some say federated) domains…
- We can (automatically) integrate disparate data sites/sources via graph merging !
- If http://hhs.gov/about returns;
- <owl:sameAs> <http:// dbpedia.org /DHHS>, <http://www. freebase.com /HHS> .
- All these data from each site is seen as one dataset, substantially lowering coordination costs of integration!
- Who is doing this in Industry?
- What about Government Agencies?
- Library of Congress Subject Headings
- http://id.loc.gov – pace setting example!
- (Homepage: ‘Linking Open Government Data’)
- Data.gov PMO SemWeb and Linked Data Team
- Collaboration with tw.rpi.edu – SemWeb inventors
- Federal CIOC-AIC Data Architecture Subcommittee Open Government Vocabulary WG
- NLM: UMLS ‘SPARQL endpoint’
- CMS: Data.gov PMO SemWeb Team members
- CMS Dashboard vocabulary WIP
- CHDI: vocabularies and URI schemes
- Rich linking scenarios to explore;
- What are successful community intervention tactics to combat childhood obesity?
- For these geographic health stats, what evidence based provider payment/performance trends emerge?
- What other Gov data is relevant? Is there something in the air or water?
- Data Web + Social = Social Data Web!
- Consider the metadata creation (domain SME’s) and instance data curation (info workers) as objects of social collaboration.
- Activity-stream history feeds, viral expert networking (etc.) will all contribute to enhanced data quality .
- Think structured data wikis, where tags are suggested and come from SME designed ontologies (vocabularies) instead of user-generated folksonomies.
- Get involved with the Data.gov PMO and/or the FCIOC-AIC-DAS-OGV-WG!
- (Come to DC Semantic Web meetups too...)