🍟Methodology

The methodology employed for this project integrates the Extreme Design (XD) methodology with a novel approach encompassing five key stages:

1. Context and state of the state-of-the-art literature

2. Large Language Models (LLMs)

3. Ontology Design Patterns

4. New Modules Integration and Alignment

This methodology reinterprets the iterative nature of XD by integrating cutting-edge LLMs, aiming to utilize them as powerful tools for ontology design.

1. Context and state-of-the-art literature

In this phase, corresponding to Task 1 of XD, designers begin by acquainting themselves with the subject matter. They first gather a broad understanding of the problem at hand, then progressively delve deeper to expand their knowledge of the specific domain of interest.

2. Large Language Models (LLMs)

At this stage of the workflow, the goal is to use the LLM to start development of Extreme Design (XD). Questions asked during this process include:

• Definition of the Studied Memes (Task 1): This step aims to integrate the knowledge gathered in the previous phase, focusing particularly on establishing and refining the terminology related to memes.

• Examples of specified meme description (Task 2): In this phase, the LLM replaces the client’s role in the XD, providing examples of meme scenarios.

• Give a user story based on one of the previously obtained scenarios (task 2)

From the ChatGPT and Claude user scenario, we derived competency questions (XD Task 4) to understand what ontology should represent and how it should do so. We then formalized these competency questions in SPARQL queries, which were then tested in our final ontologies using Protégé.

3. Ontology Design Patterns (ODP)

In this third stage of our workflow, we build on the results of the LLM modeling and the competency questions (CQs) to align them with pre-existing patterns (XD task 7). Ontology Design Patterns are employed to address common issues in ontology design and support a pattern-based approach. The designer searches for relevant ODPs in the repository, aiming to identify relationships and correlations that can be used to map the classes and properties in the developing ontology.

4. New modules integration and alignment

Finally, the newly created ontology modules are integrated into the overall structure, ensuring alignment with existing modules (XD tasks 8-10). Each module is modeled into an OWL file using Protégé.