🗒️Methodology

Steps and Methodologies adopted in the CBI Ontology Design project

This section explains the process and methodology used to create specific ontologies for various cognitive biases. The approach included several steps:

• Literature Search

• Ideation and Design

• Ontology development and modeling

• Documentation and visualization

Iterative workflow of eXtreme Design Methodology [1]

1. Literature Search

   The initial step involved an extensive literature search to gather information on the cognitive bias concept. The literature review helped in understanding the underlying theories and categorizations of cognitive biases along with providing a foundation for ontology development. This was mostly done using online sources (e.g. Wikipedia, DecisionLab) and large language models (LLMs) to extract and gain information on each bias.

2. Ideation and Design

   After the literature review, all the information obtained was collected in a Miro board to carry out the ideation phase. The next step was to start designing the ontologies. For the design and development steps, the eXtreme Design (XD) methodology was employed. It is used to identify the main requirements for each ontology through Competency Questions (CQs) and iteratively designing small ontology modules like the divide and conquer method [1]. In the ideation and design phases, LLMs were used to accomplish the initial steps of the given XD methodology flow. The outputs helped us to get to know the project context, create user stories, and extract CQs from requirements. The whole conversations with LLMs can be found under the Developed Ontologies section for each bias.

3. Ontology Development and Modeling

   The development phase involved creating detailed ontologies for each cognitive bias using Ontology Design Patterns (ODPs) and the principles of XD, in accordance with XD methodology flow. ODPs are reusable solutions for common ontology design problems. In this project, Content ODPs were particularly used as they can be easily adapted to XD methodology and can be directly reused by importing them into the ontology under development[2]. Then,

   The development process included:

   • Importing and specializing relevant ODPs according to the requirements of each context

   • Using Framester frames as an ontological resource and integrating these frames into the ontologies created

   • Creating new classes and properties as needed and ensuring they fit seamlessly with existing ones.

   • Utilizing Protégé for knowledge management and for creating and editing CBI ontologies.

4. Documentation and Visualization

   The final step involved documenting the ontology development process and visualizing the results:

   • Detailed documentation was maintained for each step using GitBook, including literature sources, design decisions, and modeling choices.

   • Visualization tools like draw.io and Graffoo were used to create UMLs of the ontologies, showing classes and their relationships.

   • All ontologies were saved in ".owl" format and hosted on GitHub, providing a referenceable URI for each module.

   This structured methodology ensured a comprehensive and flexible approach to developing ontologies for cognitive biases, facilitating collaboration and integration of individual modules into a cohesive whole.