PageRank approaches

hi all

there has been some work on applying Page Rank algorithms onto the semantic
web in the past. Most approaches seem to rank semantic web entities by
considering the probability to get to a specific entity - or node - while 
following the edges of the RDF graph.

I wonder if an approach of ranking a semantic web document by the probability
to stumble upon it while dereferencing URIs (follow your nose) is feasible or
has been tried. In this approach, every URI of every triple in a semantic
web document will be considered a link to another document (which can contain
triples and therefor further outgoing links - or not).

Maybe a problem with this idea is the notion of a "semantic web document" as 
many datasets are not centered around documents but around a triple store and 
documents are generated using DESCRIBE queries.

This approach also seems to favor ontologies but there are probably use cases
where this is good (choosing a triple subset for reasoning, for example). 
The vast number of links may also be a problem in computation.

A semantic web document rank also imposes rank measurements for single 
triples: The combined rank of all documents containing a speficic triple may 
be a credibility measure for this triple and the combined rank of the S, P 
and O documents (excluding literals and blank nodes of course) may be a 
measure of it's relevance.

As all of these ideas are quite obvious I wonder where the problems lie.

Regards,

Michael Brunnbauer

-- 
++  Michael Brunnbauer
++  netEstate GmbH
++  Geisenhausener Straße 11a
++  81379 München
++  Tel +49 89 32 19 77 80
++  Fax +49 89 32 19 77 89 
++  E-Mail brunni@netestate.de
++  http://www.netestate.de/
++
++  Sitz: München, HRB Nr.142452 (Handelsregister B München)
++  USt-IdNr. DE221033342
++  Geschäftsführer: Michael Brunnbauer, Franz Brunnbauer
++  Prokurist: Dipl. Kfm. (Univ.) Markus Hendel

Received on Friday, 11 May 2012 13:51:17 UTC