Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Ontology of Belief Diversity: A Community-Based Epistemological Approach

Tyler Fischella, Erin van Liemt, Qiuyi, Zhang

TL;DR

The paper addresses the challenge of building an inclusive ontology for belief systems to support AI fairness and safe deployment. It introduces a pragmatic, community-driven methodology centered on epistemological justification, organizing beliefs into a one-layer hierarchy of mid-level categories defined by Perception, Introspection, and Testimony. Through iterative design and two experimental tracks—term annotation and sentiment analysis with language models—it demonstrates that mid-level beliefs improve annotation agreement and can generalize to base-level beliefs for fairness testing. The work offers a foundation for more nuanced, culturally aware AI systems and outlines future work to expand ontology coverage to rituals, places, and deeper benchmarks for value alignment.

Abstract

AI applications across classification, fairness, and human interaction often implicitly require ontologies of social concepts. Constructing these well, especially when there are many relevant categories, is a controversial task but is crucial for achieving meaningful inclusivity. Here, we focus on developing a pragmatic ontology of belief systems, which is a complex and often controversial space. By iterating on our community-based design until mutual agreement is reached, we found that epistemological methods were best for categorizing the fundamental ways beliefs differ, maximally respecting our principles of inclusivity and brevity. We demonstrate our methodology's utility and interpretability via user studies in term annotation and sentiment analysis experiments for belief fairness in language models.

Ontology of Belief Diversity: A Community-Based Epistemological Approach

TL;DR

The paper addresses the challenge of building an inclusive ontology for belief systems to support AI fairness and safe deployment. It introduces a pragmatic, community-driven methodology centered on epistemological justification, organizing beliefs into a one-layer hierarchy of mid-level categories defined by Perception, Introspection, and Testimony. Through iterative design and two experimental tracks—term annotation and sentiment analysis with language models—it demonstrates that mid-level beliefs improve annotation agreement and can generalize to base-level beliefs for fairness testing. The work offers a foundation for more nuanced, culturally aware AI systems and outlines future work to expand ontology coverage to rituals, places, and deeper benchmarks for value alignment.

Abstract

AI applications across classification, fairness, and human interaction often implicitly require ontologies of social concepts. Constructing these well, especially when there are many relevant categories, is a controversial task but is crucial for achieving meaningful inclusivity. Here, we focus on developing a pragmatic ontology of belief systems, which is a complex and often controversial space. By iterating on our community-based design until mutual agreement is reached, we found that epistemological methods were best for categorizing the fundamental ways beliefs differ, maximally respecting our principles of inclusivity and brevity. We demonstrate our methodology's utility and interpretability via user studies in term annotation and sentiment analysis experiments for belief fairness in language models.
Paper Structure (13 sections, 8 figures, 3 tables)

This paper contains 13 sections, 8 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Our Contribution
  3. Previous and Related Works
  4. Epistemology and Beliefs
  5. Challenges and Limitations
  6. Benefits of Belief Ontology for AI
  7. Ontology Design and Methodology
  8. Ontology Iterations
  9. Belief Identification
  10. Experiments
  11. User Annotation Studies
  12. Sentiment Analysis for Large Language Models
  13. Conclusion

Figures (8)

  • Figure 1: Ontology of general religious thinking or worldview that is inclusive but lacks specificity. Cited from thouki2019role
  • Figure 2: First ontology attempt with multiple hierarchies and graph splits, including categorization of beliefs based on number of Gods, a derived property that is not inherent to each belief system.
  • Figure 3: Second ontology attempt that incorporates more fundamental epistemological understanding, such as dharma and religion.
  • Figure 4: Final proposed Ontology of belief systems that include epistemological justification as the main ontological hierarchy between major belief terms.
  • Figure 5: Agreement Distribution for Mid Level Ontology Nodes: Community-driven Group Participants
  • ...and 3 more figures