Shridhar Bhalerao

Document Sample
Shridhar Bhalerao Powered By Docstoc
					Finding Implicit Relations in the
         Semantic Web
          Shridhar Bhalerao
             CMSC 601
Semantic Web
 Is Web of Data .
 Each Data Resource is represented by a URI
 Data resources are linked to each other .The link also
  has a URI identifying it.
 Ontologies are used for conceptualizing , data is
  mapped to the concept.
 Described Using RDF.
Ontologies, RDF Triple, isparql
 Conceptualization of data into classes.
DBpedia Ontology .

 Properties , Domain , Range.

 RDF.

 Example
Querying the semantic web
 Queries
  The queries are in the form of Rdf graph.

 Isparql
A tool for querying RDF datasets .

User must know properties and classes of the ontology
 the dataset is using .
Problem Statement
 User query given in a single triple but the answer is present
   as n triples .

 The relation is not present directly in the database.

 The user creates the relation which he thinks is appropriate.
          works for

Actor               director
 Finding the N triple form of the user query .

 Algorithm for finding destination class which scan all
  the properties in the ontology that have the domain of
  the 1st class .Create a triple with candidate class.

 Tree form.

 Output will give the links from the 1st class to
  destination class.
 Generate a set of queries having implicit relations from
  users .

 Generate the correct triple from an ontology expert.
  Retrieve the answer set.

 Find the correct triple using the algorithm.

 Evaluate the result using precision and recall.

 Presence of Multiple Paths.
   The two classes can be related by different relations .

 Capturing the semantics of the relation given by the
  user and using it to find the correct path.

Shared By: