Swoogle : A Semantic Web Search Engine

1
Swoogle A Semantic Web Search Engine
2
Some Facts

• Developed at and hosted by UMBC
• Crawler based, claims to have indexed over 1.3
  million Semantic Web Documents (SWDs)
• Users can search documents, terms
• Provides Interface for users as well as agents

3
Architecture
4
Harvesting

• Uses Google APIs to generate candidate URIs,
  since the Semantic Web Graph is very sparse now
• Uses own crawlers using these as seeds
• Allows users to submit URIs

5
Metadata Generation

• Used for ranking, falls in three categories
• language feature Basic, RDF statistics and
  ontology annotation.
• RDF Statistics Tells you whether it is a SWD or
  SWO
• Ontology Annotation describes Antologies

6
Making use of Semantics
7
Ranking Pages The Rational Surfer

• Picks a SWD randomly
• Either terminates surfing or picks a node in RDF
  Graph depending upon term frequency
• Either surfs to another document , terminates
  surfing, or picks another node depending on what
  node it has chosen

8
Calculating PageRanks

• Captures the idea that links followed from a page
  depend
• on the correlation between two in terms of
  links
• Notice that raw PageRank of x is multiplied by
  the weight
• of links between x and a and the weight of all
  outgoing
• links from a.
• SWDs and SWOs have different formulae for
  Pageranks.

9
The Real Search Process
10
References

• Swoogle A Semantic Web Search and Metadata
  Engine Li Ding et. al. http//ebiquity.umbc.ed
  u/get/a/publication/115.pdf
• Search Engines for Semantic Web Knowledge - Tim
  Finin, Li Ding - http//ebiquity.umbc.edu/paper/ht
  ml/id/304/Search-Engines-for-Semantic-Web-Knowledg
  e
• Search on the Semantic Web Li Ding et. al. -
  http//ebiquity.umbc.edu/paper/html/id/257/
• Swoogle A Search and Metadata Engine for the
  Semantic
• Web Tim Finin et. al. -
  http//ebiquity.umbc.edu/paper/html/id/182/

11
Questions / Comments?

• THANK YOU!