W3C home > Mailing lists > Public > semantic-web@w3.org > March 2005

Re: Frequency List of resources and properties

From: Yuzhong Qu <yzqu@seu.edu.cn>
Date: Wed, 30 Mar 2005 15:16:35 +0800
Message-ID: <005b01c534f8$68d7bcf0$fd0b77ca@xobjects>
To: "Jeremy Wong" <50263336@student.cityu.edu.hk>
Cc: "SWIG" <semantic-web@w3.org>
I agree with you that there can be different statistics for different purposes.

So some statistical models should be established before a large amount of ontologies are collected and analysed. 

As to the merging issue, there are at least two ways to manage it. One is to record the "owl:imports" linkage between ontologies (BTW, there may exist many versions for one ontology), another way is to sum up the statistics. 

Please note: If the infered axioms (by RDFS or OWL inference rules) are included in statistics, then the computing will become too complex.

OK, I can see that the indexing issue is related to the statistics.


Yuzhong Qu
  ----- Original Message ----- 
  From: Jeremy Wong 
  To: Yuzhong Qu 
  Cc: SWIG 
  Sent: Wednesday, March 30, 2005 12:48 PM
  Subject: Re: Frequency List of resources and properties


  Your intuition applies on RDF schema without entailments. It seems that there can be different statistics for different purpose. For examples, a statistics for RDF in general, a statistics for RDFS, for OWL, and statistics for various ontologies, RDF closures.

  It is likely that there can be a statistics for each ontology. When ontologies are merged together, a merge of the statistics of the ontologies may be useful. However, how to merge statistics?

  BTW, my objective to use the statistics is for indexing triggers, providing that there is a pool of rules for real-time inference.


  Jeremy
    ----- Original Message ----- 
    From: Yuzhong Qu 
    To: Jeremy Wong 
    Cc: SWIG 
    Sent: Wednesday, March 30, 2005 10:24 AM
    Subject: Re: Frequency List of resources and properties


    This is one of the works planned  in our Group. Hopefully, the statistics will be given around Oct. this year.

    Frankly,  your list is not fit with my intuition.

    Intuitively, the following classes and properties have high frequency of use

    rdfs:Class
    rdf:Property

    rdfs:subClassOf
    rdfs:domain 
    rdfs:range
    rdfs:subPropertyOf


    BTW,  the statistics model ( and objectives) is not easy to established,  and the ontoloy capturing  is also a hard stuff.  


    Any comment is welcome!


    Yuzhong Qu

      ----- Original Message ----- 
      From: Jeremy Wong 
      To: semantic-web@w3.org 
      Sent: Wednesday, March 30, 2005 2:50 AM
      Subject: Frequency List of resources and properties


      I want to make a list of resources and a list of properties, order by frequency of use. It would be grateful if you can help on improving the lists and even add new item to the lists.

      A list of resource (exclude properties):

      -- higher frequency
      rdfs:Resource
      rdf:Property
      rdfs:Class
      rdfs:Literal
      rdf:List

      -- lower frequency
      rdf:Alt
      rdf:Bag
      rdf:Statement
      rdf:Seq
      rdf:XMLLiteral
      rdfs:Container
      rdfs:ContainerMembershipProperty
      rdfs:Datatype
      owl:AllDifferent
      owl:AnnotationProperty
      owl:Class
      owl:DataRange
      owl:DatatypeProperty
      owl:DeprecatedClass
      owl:DeprecatedProperty
      owl:FunctionalProperty
      owl:InverseFunctionalProperty
      owl:Nothing
      owl:ObjectProperty
      owl:Ontology
      owl:OntologyProperty
      owl:Restriction
      owl:SymmetricProperty
      owl:Thing
      owl:TransitiveProperty

      A list of properties:

      -- higher frequency
      rdf:type
      owl:sameAs
      rdf:first
      rdf:rest
      rdf:nil
      rdfs:domain
      rdfs:range
      rdfs:member
      rdfs:subPropertyOf
      rdfs:subClassOf

      -- lower frequency
      rdf:object
      rdf:predicate
      rdf:subject
      rdf:value
      rdfs:comment
      rdfs:isDefinedBy
      rdfs:label
      rdfs:seeAlso
      owl:allValuesFrom
      owl:backwardCompatibleWith
      owl:cardinality
      owl:complementOf
      owl:differentFrom
      owl:disjointWith
      owl:distinctMembers
      owl:equivalentClass
      owl:equivalentProperty
      owl:hasValue
      owl:imports
      owl:incompatibleWith
      owl:intersectionOf
      owl:inverseOf
      owl:maxCardinality
      owl:minCardinality
      owl:oneOf
      owl:onProperty
      owl:priorVersion
      owl:someValuesFrom
      owl:unionOf
      owl:versionInfo

      Frankly speaking, I don't have much statistics data and I am even lazy to produce one (my time is too tight.. my priority may not allow so.. work->study->work->study->... -_- and I am not too smart -_-...). The above 2 lists are my prediction only, modification with ground is totally welcome.


      Jeremy
Received on Wednesday, 30 March 2005 07:13:59 UTC

This archive was generated by hypermail 2.4.0 : Tuesday, 5 July 2022 08:44:52 UTC