A Framework of Change Operators for Customisable Ontology Evolution:

Change operators are the building blocks of ontology evolution. Different layers of change operators have been suggested. In this work, we proposed an approach to deal with ontology evolution, in particular, change representation as a pattern-based layered operator framework . As a result of an empirical study, we identify four different levels of change operators based on the granularity , domain-specificity and abstraction of changes. The first two layers are based on generic structural change operators, whereas the next two layers are domain-specific change patterns . These layers of change patterns capture the real changes in the selected domains.

* The work has been published in peer reviewed conference (OntoContent 2009).

A Layered Framework of Change Logs for Pattern-Based Ontology Evolution:

In this work, we proposed a layered framework for ontology change representation. The framework maintains the operationalisation of manipulation support through layered change logs . Layered change logs capture the objective of the ontology changes at a higher level of granularity and helps in comprehensible understanding of ontology evolution. The layered change logs are formalised using a graph-based approach . In order to conceptualise the ontology changes, we constructed a metadata ontology by looking into concrete structure of OWL DL syntax-based domain ontologies. Sesame native triple store has been used for storage of the domain ontologies, metadata ontology and change logs. The change log and ontologies are recorded in form of RDF triples in a sesame triple store repository. SPARQL format queries are used to extract the data from the triple store repository. Our ontology editing framework (OnE) offers an graph API which can be used for i) generating graphs (of type GraphML) from change log triples, extracted from sesame repository and ii) reading graphs. Our next step is writing the algorithms for the identification of frequent change patterns.

* The work has been published in peer reviewed conference (ODISE 2011).

Graph-Based Ontology Change Pattern Discovery:

Change patterns can provide guidelines to content change management systems and support in evolution process of the software system components. While patterns are sometime used in their exact form, often more flexibility is needed. To capture semantically equivalent, but operationally different patterns, we introduced a metric called sequence gap or node-distance . On the basis of our empirical study of ontology change log graph, we identified the recurrent ontology change patterns of two types, i.e. Ordered Change Patterns (OCP) and Unordered Change Patterns (OCP) . Such patterns are not only suitable to provide pattern level change support for particular domain ontology but can also be shared among ontologies that have similar conceptualization. The performance comparison between OCP and UCP algorithms can be found here.

* The work has been published in a peer reviewed conference (ISWC: EvoDyn 2011).

Ontology-based Modelling for Consistent Content Change Management:

In this work, we proposed an approach to deal with model-driven content-centric information systems and access to their content. At the core of our approach is an ontology-based semantic annotation technique for diversely formatted content that can improve the accuracy of access and systems evolution. Domain ontologies represent domain-specific concepts and conform to metamodels . Different ontologies - from application domain ontologies to software ontologies - capture and model the different properties and perspectives on a software content unit. Interdependencies between domain ontologies, the artifacts and the content are captured through a trace model . The annotation traces are formalised and a graph-based system is selected for the representation of the annotation traces.

* This work has been published in a peer reviewed conference (ICOSE 2010)