Unit 6: Introduction to Ontology Building and Online and Offline Tools

Overview

As per the course website, "This unit focuses on the practical application of ontological approaches for the development of a knowledge-based system. We will look to apply the knowledge gained so far to aid this development via a case study."

My Reflection

Overall Reflection

This unit aimed to tackle two aspects: First, methodologies to build ontologies, and second, introduction to protege software by Stanford University, which is an open-source ontology editor and knowledge management system. The readings were a mix of guides on how to use the software, and a research paper by Demoly, Kim and Horváth (2019) referring to two different ontology building methodologies. The formative activity was related to the methodologies, while the software introduction would be essential for the modelling assignment that is due in Unit 11.

Artefacts

Formative Activity: Reflection on Demoly, Kim and Horváth (2019)

The formative activity invited us to reflect on the article titled Ontological engineering for supporting semantic reasoning in design: deriving models based on ontologies for supporting engineering design' by Reflection on Demoly, Kim and Horváth (2019), through the following three questions and prompts:

Reflect on the different strategies for developing ontologies.

The paper explicitly mentions two different strategies for developing ontologies, which are: top-down and bottom-up approaches. The top-down approach starts with high-level abstract concepts, progressively refining them into domain-specific classes and relations. Meanwhile, the bottom-up approach begins with specific data, instances or use cases, then generalises into broader concepts.

In addition to these two methodologies, there are other ones that are not explicitly mentioned in the article, like the middle-out approach that starts with the mid-level most notable concepts, and then expands upwards and downards. There is also the reuse approach, which is to build on existing ontologies and adapt them to new domains or applications.

What are the benefits and drawbacks of each approach?

The top-down approach benefits are that is ensures consistency and coherence in the ontology structure. It also facilitates interoperability between systems, and even paves the way for standardisation and reusability across different domains. However, its drawbacks include being more time-comsuming and resources intensive, and its holistic conceptualisation may make it harder to apply.

The bottom-up approach benefits are that it is quicker and more direct to apply in specific use cases, and it also ensures capturing domain or applicaiton specific knowledge. Meanwhile, its drawbacks that it may not scale for integration and interoperability between systems, and leads to fragmented knowledge representation that is isolated or hard to incorporate into ontologies of other domains.

Consider various scenarios and which methods would be suitable.

For its holistic approach, the top-down methodology would be more suitable for large-scale and wide-perspective projects, especially that is similar to founding frameworks or standards, as in establishing an aerospace industry design framework, or a healthcare service provision framework that can be used across different hospitals and clinics.

On the other hand, the bottom-up approach would be more suitable for direct applications or specific projects, like establishing a new assembly line of a factory, or planning a new product portfolio (menue) for a restaurant.

Artefacts

Useful Additional Resources

3.7 How to Design your own Ontology

I found this video and its related playlist more helpful to understand the construction of ontologies and how protege can help in this regard, more than the practical guide readings.

Reference List

Demoly, F., Kim, K.-Y. and Horváth, I. (2019) ‘Ontological Engineering for Supporting Semantic Reasoning in design: Deriving Models Based on Ontologies for Supporting Engineering Design’, Journal of Engineering Design, 30(10-12), pp. 405–416. Available at: https://doi.org/10.1080/09544828.2019.1633626.