A Comparison of Description Mechanisms for URI Collections
Tony Hammond, 2008-12-06
(updated from previous version: 2008-12-04)
  ORE POWDER SITEMAP
Name Object Reuse and Exchange Protocol for Web Description Resources Sitemaps
Sponsor OAI (Open Archives Initiative) W3C sitemaps.org (supported by Google, Yahoo!, and Microsoft)
Home Page http://www.openarchives.org/ore/ http://www.w3.org/2007/powder/ http://www.sitemaps.org/
Status Published Specs (1.0) and User Guides
http://www.openarchives.org/ore/1.0/ 		(In Development) Last Call Working Drafts, Nov. '08
http://www.w3.org/2007/powder/ 		Published Spec (0.90)
http://www.sitemaps.org/protocol.php
Timeline Oct. '06–Oct. '08

Oct. '08, public release (1.0), Feb. '08, beta release (0.9), Dec. '07, alpha release (0.1) 		Feb. '07–

W3C CR status expected '08/'09, and W3C Rec early '09 - note that the WG was scheduled to complete Mar. '08 		Jun. '05–

Google introduced (0.84) in Jun. '05, with Google, MSN and Yahoo announcing joint support for (0.90) in Nov. '06
Primer yes yes no (not required)
Documents
Aim "data model for the description and exchange of aggregations of Web resources, named Aggregations" "a mechanism through which structured metadata ("Description Resources") can be authenticated and applied to groups of Web resources" "Sitemaps are an easy way for webmasters to inform search engines about pages on their sites that are available for crawling"
Key terms "aggregation", "aggregated resources", "aggregation graph", "resource map" (also "compound digital object") "description resource", "resource set", "descriptor set", "tag set" "sitemap", "sitemap index"
Audience digital repository, content publisher general webmaster
Concerns aggregation boundaries, compound objects, digital repository, object exchange operational semantics (XML), formal semantics (OWL), pattern-based URI sets, simple properties and/or free tags an XML-based URI listing
Ease of use low medium high
Learn curve high¹

¹ classes as "high" on account of its abstract data model (with no fixed bindings - although three bindings are defined), and its introduction of the concept of a named "aggegation" resource 		medium/high¹

¹ POWDER with its operational semantics (XML) would class as "medium", whereas POWDER with formal semantics (RDF/OWL or POWDER-S) would class as "high" 		low
RDF-based yes yes no¹

¹ this sitemap extension from DERI (Nov. '07) allows site-based RDF data to be identifed using a sitemap
Comments presents abstract data model which can be serialized variously – guidelines given for Atom, RDFa, RDF/XML bindings – considerable emphasis placed on discovery methods – documentation dwells on web architecture principles presents data model with two fixed bindings: XML and RDF/OWL – can be used as simple XML application with no special knowledge of RDF required presents data model encoded as a simple XML schema - limited to describing single site - constraints on size of sitemaps and sitemap indexes
Serializations unfixed: Atom, RDFa, RDF/XML, ... fixed: XML (RDF/OWL) fixed: XML
Discovery multiple methods (see ORE User Guide Resource Map Discovery), also resource map URI/aggregagation URI binding mechanisms multiple methods (see POWDER Description Resources, Sect. 4: Associating Resources and DRs) publication, form submission, robots.txt, HTTP (see Sitemaps Protocol: Informing search engine crawlers)
USP identifies aggregations resource grouping patterns simple inventory listing
Pros defines compound objects powerful pattern matching for URI sets, inclusions and exclusions, uses IRIs  
Cons requires HTTP¹ URI for aggregation URI (since uses HTTP machinery for discovery purposes), no exclusions

¹ spec talks of "protocol-based" URIs (i.e. URIs based on a network protocol) - though generally taken to mean HTTP URIs 		no relative URIs, no blank nodes, no datatypes, only simple (literal, resource) objects, requires semantic extension to OWL limited to URLs from single site, no relative URIs, size constraints on sitemaps (50,000 URLs; 10MB) and sitemap indexes (1000 sitemaps; 10MB)
URI patterns no yes yes
Requirements the resource map must:
  • specify the resource map URI
  • specify one (or more) aggregation URIs
  • include a ore:describes property relating resource map URI to aggregation URI
  • include a ore:isDescribedBy property relating aggregation URI to resource map URI
  • include a dcterms:creator property
  • include a dcterms:modified property
the description resource¹ must:
  • begin with the root element <powder>
  • include a child element <attribution>
  • the <attribution> element must include one <issuedBy> element
  • the <attribution> element may include one <issued> element
  • include zero (or more) <dr> description resource elements
  • each <dr> must contain at least one non-empty <iriset> element
  • each <dr> may contain at least one non-empty <descriptorset> element (and/or <tagset>) that may contain RDF/XML describing the IRIs in the <iriset>
¹ description resource encoded as a POWDER document (rather than a POWDER-S document) 		the sitemap must:
  • begin with an opening <urlset> tag and end with a closing </urlset> tag
  • specify the namespace (protocol standard) within the <urlset> tag
  • include a <url> entry for each URL, as a parent XML tag
  • include a <loc> child entry for each <url> parent tag
note that all other tags are optional
also, all URLs in a sitemap must be from a single host, such as www.example.com
Extensions open RDF-based model custom URI sets, new properties via XMLNS new properties via XMLNS¹

¹ as an example see this sitemap extension from DERI (Nov. '07) which allows site-based RDF data to be identifed
Tools Atom Resource Map Validator
see also: Tools and Additional Resources 		POWDER Validator, POWDER Processor
see also: Tools 		Sitemap generator
see also: Third-party sitemap tools
see also: XML-Sitemaps.com