Realizing the Semantic Web using XML and OWL ontologies

 

Ph D Thesis, Sujatha R Upadhyaya, July 2007

 

In the Semantic Web, the emphasis lies on associating semantics with data.

This process, often known as the semantic annotation of data, is the starting

point for making the web applications inter-operate. Semantic interoperability

falls in place, if the data on the web is described using standard vocabulary.

Ontologies come into this picture as the sources of such standard

vocabulary. The other task expected to be accomplished with ontologies

in the context of the Semantic Web is that, they must form the logical

reasoning support for deriving inferences from data.

As of today, we have few standard domain ontologies that can be employed

for annotating data on the web. Keeping this mind, this thesis examines

the methodologies for building domain ontologies, for employing them in

semantic annotation of data and for building systems that can reason with

ontologies. We accomplish each of these tasks by building appropriate

tools that help web applications to migrate to the Semantic Web and bene-

fit from logical reasoning. Our methodologies are particularly applicable to

the structured web content that appears as XML Databases and not to the

kind of data that appears as plain text. We represent ontologies in OWL,

Web Ontology Language, the only recommended standard is for ontology

representation.

 

To begin with, this thesis investigates the means for building ontologies

to provide standard vocabulary for describing the data on the web. For

marking an advancement in attempts to build ontologies from structured

data, this thesis proposes a method of building ontologies from Entity

Relationship models and Extended Entity Relation models. It discusses, also

the implementation of a tool named ERONTO that builds OWL ontologies

from E/R and EE/R models. The result is particularly significant as E/R

and EE/R form the richer source of domain knowledge compared to relational

schema and data, which were used in previous attempts.

Employing ontologies effectively for annotating one’s data, is rather difficult

because of its complexity of representation. In this direction, this thesis explores

the methods of making ontologies easily employable by XML users.

We proposed a procedure for presenting the vocabulary in OWL ontologies

in XML Schema. With this procedure, we built a tool called OntoSchema.

The output schema of OntoSchema is particularly helpful in choosing the

right term in the right context for describing one’s data. With this effort,

we studied the expressiveness of two language standards namely OWL and

XML schema and suggested the limit of expressiveness of XML Schema.

Reasoners form the inference engines in the Semantic Web. Querying with

a reasoner could be time intensive especially with large ontologies. This

thesis examines the possibilities making quick inferences by storing the

ontology in a suitable form. We built a data structure called Knowledge

Table, which intuitively stores the constraints on properties as applicable

to concept descriptions. With Knowledge Table, typical reasoning tasks are

reduced to table look-ups. It provides a mechanism for interpreting the semantics

of OWL constructs. However, we restrict the expressiveness of the

language by specifying certain rules. We demonstrate the use of Knowledge

table for identifying the instances of ontology concepts, represented as XML

data. We also demonstrate its use in enhancing ontologies by adding new

concepts. By employing the knowledge table in instance identification and

ontology enhancement we illustrate how to perform the general reasoning

tasks such as, checking subsumption, disjointness, satisfiability and finding

the most specific concept of an individual.