interpretri

Interpretri is a framework for interpretable triadic reasoning in data analysis, centered on three-way interactions among entities and on producing explanations that humans can understand. It is used for modeling and explaining decisions that depend on triads rather than pairwise relations alone.

Origin and name: The term interpretri blends “interpret” and “triad.” It originated in discussions of explainable AI and network analysis to address the need for transparent reasoning about triadic patterns in complex data.

Core concepts and components: Triadic representation encodes three-way relations, often via triadic motifs or relational tensors. The interpretable component generates human-readable explanations—rules, feature contributions, or visual summaries—that attribute a model’s triadic decision to a small set of factors. Evaluation uses fidelity to the triadic model, cognitive simplicity, and coverage of important triads.

Applications: Social network analysis to understand how triads influence link formation; knowledge graphs and recommender systems where triadic context informs predictions; biology and chemistry where three-molecule interactions impact outcomes. Example: in a collaboration network, interpretri might explain why a triad of researchers forms a collaboration based on individual attributes and mutual connections.

Limitations and relation to other work: Triadic analysis is computationally intensive for large datasets, and explanations may be approximations. Interpretri relates to broader fields of interpretable machine learning, network motifs, and explainable AI.