Linda Hill: NKOS intro
Digital libraries focused on collections
Reference works a superset of gazetteers
Geographic place: name, type, footprint: no shared typing schemes
Distributed search vs. aggregation
(Someone) asked about ranking places with same name, e.g. Paris, France, is most important; can be seen as prior probability
Goal of UCSB is document->footprint, not just name->location
ECAI not building gazetteer, but looking at specifying needs of cultural community
Content standards, feature type thesaurus, best practice guidelines
Relationship types should be standardized
http://www.alexandria.ucsb.edu/gazetteer/protocol
http://www.alexandria.ucsb.edu/thesaurus/protocol
What about selecting default relationships, e.g. administratively part of or spatially part of?
Should this mapping be in server or client?
Should thesaurus protocol allow multiple relation types?
Yes!
Filter encoding in; XML out
Use of XLink to reduce redundancy?
http://www.giub.uni-bonn.de/gisfe/
How to unify ADL and OpenGIS approaches?
Difference between web gazetteer service and web feature service?
NKOS:
Standards phrased in terms more relevant to non-lexicographers
Registry for DC Subject schemes
Danger of NISO thesaurus standard
Suggested that British standards bodies get coordinated
Academia Sinica:
Name authority, etymology, history, etc.
http://www.sinica.tw/~gaz…
geotools open source visualization toolkit
http://www.gbhgis.org
http://www.geotools.org
Microsoft Knowledge Network Group
GNIS names often don’t exist
http://atlas,gc.ca
Cliff Lynch’s response:
Concessive clause: not expert, etc.
Reaction to things he didn’t hear
Nature of gaz.:
Minimalist view: nuisance to real geography, unfortunate effect of language (early-mid 1990s)
Middle view: not just geometric inclusion but more conceptual (polit., legal) space
Creep to skeleton of all knowledge
Services should not be encyclopedias
Not just “:find a place”. More complex structuring for DLs: history, culture, etc.
Nasty engineering problems: persistent Ids when referencing texts
Distrubed gaz. Issues: scoping, provenance, versioning
Peter Buneman of Penn on data provenance
Issue of gaz scoping: aspiration to reconstruct the past, e.g. architecture
Do we want all Main Streets in our gazetteers?
Language of place names?
Spoken place names
Concern about combining update with retrieval
We need more agreement in order to update than to retrieve
How much context to carry with query
Cf. Experience of Z39.50
Moving elaborate semantics can be tricky
e.g. speed in implementation vs. usefulness
Standards issues:
ADL vs. ISO
Supersession of theoretic standards by tinkering
Standards vs. services
Who will put up data people want with these standards?
Killer app will define standard
No discussion of how to make case for this stuff
Simple visualization of documents is mere party trick
Geography a way of finding associated concepts and doing query expansion
Proposals for extra subject headings
How to improve scholarship in demonstrable ways?
Data quality presentation was intriguing:
NSDL offers opportunity for local gazetteers driven by education
Discussion:
Delaware has DataMill. People can input info. State gov and univ are involved under USGS
USGS rules for places vs. popular names
Scope restrictions due to Homeland Security?
Signposts down in Britain, Soviet and Franco-era maps
Power, water, infrastructure maps
Locations of endangered species
Restrictions w/o rational threat assessment
Archaeologists may intentionally fuzz data
Related to privacy problems with public record data
Feature types are key for connecting geographic query to broader IR
We shouldn’t develop thesauri lightly
And we can have multiple feature schemes
Different schemes may break down farther. Go from leaf in general thesaurus to more specialized thesaurus
Comparison to DC
Focus on definitions for relationships; faceting
But combining different thesauri is hard
Even ADL may to too complex
Lex Berman wants only 10 basic feature types
Scoping problem: Categories of British admin units only applies to Britain
Are administrative units too Western?
Commercial uses?
ESRI is providing place finder for free up to 5000 hits a day.
Distinction between place names and addresses: geoparsing vs. geocoding
Linda:How does georeferenced info fit into CNI program lines?
Cliff: Increasing interest in structured reference resources. As scholars produced more reference works, we need structure