TWFY RDF Mapping
From GetSemantic
For further development, see: the semwebparlparse kforge project we've set up.
Anybody interested in helping figuring out an RDF mapping for TheyWorkForYou ?
The XML is available here, and it wouldn't be hard to write XSLT to RDF/XML (I've done some preliminary XSLT already).
We'd just need to decide what ontologies to use, and what URIs to use for MPs, Parties, Constituencies, etc (it's mostly possible to use the fragment identifiers in the XML files, but perhaps wikipedia pages for the other entities?).
If you don't know RDF, don't worry, you can still help with the mapping - it's like OOP. Everything is a type of Class, and has properties. Property values will either be literals, or another class instance.
Contents |
[edit] Why bother?
For semweb enthusiasts, I think this is ripe low hanging fruit. The TWFY folks have done the hard work of scraping the data - an RDF mapping would not be much work at all, and can show-case RDF's capabilities nicely - it's flexibility and mergebility really shine here.
For people interested in the political data, RDF technologies have a lot to offer - SPARQL would provide a more powerful query interface than currently available, and a far broader range of information can be incorporated into the dataset. Calendars of appointments and meetings, statistical information about constituencies, publications by and about politicians and parties, related news items. We can facilitate an incredibly rich and useful data-set here (and the data provided by TWFY is already pretty darn impressive and useful).
For interface designers, the challenge of giving non-SPARQL experts access to the power of an RDF version is a good opportunity to demonstrate your ingenuity ;)
[edit] Possibly useful ontologies
- FOAF
- Politico (see also various other political ontologies in this directory here: http://www.govtrack.us/share/ and see sample data here: http://www.govtrack.us/data/rdf/ )
- vCard (for contact details ?)
- geo (for geo location of constituencies etc)
- bibtex (for publications by politicians)
- dc
- time (for dates ?)
[edit] Proposed Classes
- Politician (pol:Politican - subclass of foaf:Person)
- Office (pol:Office? - should we subclass this? ) - the political office held by a politician
- Constituency (a subclass of pol:Organization ?)
- Party (a subclass of pol:Organisation)
- RegisteredInterest (eg: a declared extra source of income))
- We ought to get the Register of Member's Interests and find a way of codifying it.
- Debate
- Speech
- Report
- Question
- Answer
[edit] Proposed properties
- pol:hasRole
- pol:represents
- pol:forOffice
- twfy:fromWhy (the reason why they are in office eg: general election).
- twfy:toWhy
[edit] Pre-existing properties with class relations
Most of the FOAF and/or vCard properties can be used for Politician.
[edit] Potential mashup applications
- DbPedia for Wikipedia data
- Freebase
- FOAFCorp : Corporate Friends of Friends
- RDF Book Mashup
- (via scraping perhaps) Blogs and websites of politicians and parties
[edit] Design issues
- If the party is 'Ind', I am currently ignoring it. Independent basically seems to mean, semantically, both "not an established party" and "a person elected without a party" and "a party so small that it's not worth mentioning". Therefore, it seems sensible not to declare a party loyalty. "That about which we cannot speak, we pass over in silence." (Wittgenstein) --Tom 14:49, 18 April 2007 (EDT)
- Should we have properties that are inverse of other properties? I am inclined to think not, given recent comments on the SIOC-DEV ml about problems caused by allowing this in SIOC
