Zubeida Khan    (3JMH)

Zubeida Casmod Dawood
PhD Student
University of KwaZulu-Natal
Westville, South Africa.

email:  zkhan [at] csir.co.za    (3JMI)

My Supervisor is Dr. MariaKeet    (3JML)

MSc dissertation: Foundational Ontology Interchangeability with the Repository of Ontologies for MULtiple USes (ROMULUS)    (3WES)

April 2012 - July 2013    (3WEP)

The notion behind foundational ontologies is to have a single foundational ontology to serve as a basis for providing high-level entities and relations that are common between all ontologies in order to facilitate interoperability among heterogeneous systems. However foundational ontologies alone do not suffice in solving the problem of interoperability, due to the fact that many foundational ontologies exist, each with conflicting philosophies. The WonderWeb Foundational Ontologies Library (WFOL) was envisioned to facilitate interoperability, but not implemented, possibly due to a lack of: ontology mediation (alignment, mapping and merging) techniques, documentation and comparisons between foundational ontologies and modularisation techniques. In order to solve this problem, three widely used foundational ontologies: DOLCE, BFO and GFO were selected and a web-based repository, ROMULUS was created. Ontology mediation was performed to assist in achieving foundational ontology interchangeability between the selected foundational ontologies. Modularity was performed to simplify ontologies in order to easily perform mediation and to create modules for specific functions. ROMULUS provides the user with access to: new foundational ontology modules, mappings between foundational ontologies, merged foundational ontologies, a higher level foundational ontology containing only the most general entities common to the three foundational ontologies and a method to assist the user with performing foundational ontology interchangeability. The new modules in ROMULUS (separate endurant/perdurant modules, OWL 2 profile modules, and more/less-detailed ontology modules) are useful when one wants to perform functionality specific to the module type. The mapping and merged ontologies, which may be used together with the method for performing foundational ontology interchangeability, allow a user to convert between the three foundational ontologies and to link an ontology using a particular foundational ontology to a different ontology that uses another foundational ontology, thereby achieving transparency. The higher level foundational ontology may assist in interoperability because it is a single ontology that encompasses entities that are common between all three foundational ontologies. ROMULUS has been evaluated in terms of its foundational ontology interchangeability, accuracy of alignments and by comparing it to other repositories. From the evaluations, we realised the following: While barely 50% of the participants agreed with the alignments, real disagreement was less than 10%; foundational ontology interchangeability may be achieved using the merged ontologies; ROMULUS offers advanced functionality for most criteria when compared to other repositories. Therefore there is reason to believe that ROMULUS does assist with foundational ontology interoperability.    (3WEQ)

PhD proposal: Creating and maintaining useful modules for ontology development    (3WER)

I am currently working on ontology modularity and modular ontology evolution.    (3JMJ)

Ontologies have grown to be large and complex to the point where it causes difficulty for humans, in understanding and maintaining, and for machines, for processing and reasoning. Naturally, omitting important knowledge from an ontology in order to reduce it will affect its application and is not a solution. Modularity, however, deals with simplifying an ontology for a particular context or by structure into smaller ontologies, thereby preserving the knowledge. There are a number of benefits in modularising an ontology including simplified maintenance and machine processing, as well as collaborative efforts whereby work can be shared among experts. Modularity has been successfully applied to a number of different ontologies to improve usability and assist with complexity. However, the problem of modularity, in terms of module extraction for satisfying all purposes of modularity such as disjoint modules, modules of context, modules of greater/less details, and in terms of the quality of the generated modules in terms of size and intra-module distance, has not been satisfactorily addressed.    (3VKV)

In order to successfully solve the problem, a number of theoretical aspects have to be investigated. Foremost, one must identify properties that a `good' or `usable' module meets. It is important to determine which ontology module types are the most widely-used, and to characterise each such type by distinguishing properties. Thereafter, one must determine which ontology languages and approaches can be used to link these modules. While there are a few existing modular languages, each one compromises on specific aspects of the ontology modules, such that functions such as foreign linking and properties such as subsumption and transitivity are affected. Therefore it is necessary to investigate how to use existing resources for effective module linking such that the functions and properties of the modules are not compromised. We expect to obtain an effective method using existing, and possibly new resources for module linking and interaction. Once these problems are investigated, it is possible to design approaches or methods to assist in creating useful modules for different use-cases. Evaluating these approaches, against existing approaches, and with use-cases will assist in determining the usefulness of our approach, and the quality of the generated modules.    (3VKW)

However, these tasks alone do not solve the entire problem of effective ontology usage by modules. Another issue concerning ontology modules is the fact that ontologies are constantly changing, and since modules are interrelated, changes made to the ontology or its sub-modules tend to affect other modules. Of course, these changes must be conveyed to the entire system in order to maintain consistency. However, this is a cumbersome and timely task, which is often prone to error and omission. Presently there are no existing techniques or approaches in the field of modular ontology evolution thus we are required to investigate and design methods for this. One such method is the notion of a module management system. Presently, there is no existing system available to assist with module management, and we aim to create such a system. An evaluation of such a system is also required. We plan to evaluate the proposed system by testing it against its functional requirements, with users, and by using it to manage modules. We hope to investigate and solve the shortcomings of the modularity problem, such as satisfying different purposes of modularity, improved module quality and modular ontology evolution.    (3VKU)