Strength Change Explanations in Quantitative Argumentation

2026-03-03 · Source: cs.MA updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning · Depth: Expert, medium

Summary

This paper introduces "strength change explanations" (SXs) for quantitative bipolar argumentation graphs (QBAGs), aiming to make AI inference contestable. SXs describe specific changes to the initial strengths of a subset of arguments within a QBAG that can achieve a desired ordering based on the final strengths of another subset of arguments. The authors demonstrate that existing inverse and counterfactual problems can be reduced to SXs. They prove soundness and completeness properties, and show existence and non-existence in special cases. Using a heuristic search, the research successfully finds SXs for layered graphs common in applications, though limitations exist where guarantees for explanation presence or absence are not provided. An example illustrates how modifying initial strengths of arguments like $\mathsf{a}$ and $\mathsf{e}$ can flip a credit application decision from rejected to accepted.

Key takeaway

For AI Researchers developing contestable AI systems, understanding Strength Change Explanations (SXs) is crucial. Your work should focus on implementing and refining heuristic search methods for identifying optimal or $\epsilon$-approximate SXs in complex QBAGs, especially for real-world applications like financial decision-making, to empower users to challenge and understand AI outcomes effectively. Consider the trade-offs between explanation optimality and computational cost.

Key insights

Strength change explanations enable contestable AI by identifying minimal initial strength adjustments to alter argument outcomes.

Principles

• Contestability requires explaining how to achieve desired outcomes.
• Gradual semantics update argument weights based on graph topology.
• Optimal SXs minimize the sum of initial strength changes.

Method

SXs are found by generalizing the inverse problem in quantitative argumentation, defining (mutable argument, initial strength)-tuples that yield a desired final strength ordering, and applying heuristic search for layered QBAGs.

In practice

• Apply SXs to explain credit application decisions.
• Use QBAGs for explainable image recognition.
• Integrate SXs into recommendation systems.

Topics

• Formal Argumentation
• Explainable AI
• Quantitative Argumentation
• Strength Change Explanations
• Contestability

Best for: AI Researcher, AI Scientist, Research Scientist

Related on AIssential

• On Strong Equivalence Notions in Logic Programming and Abstract Argumentation — A new strong equivalence notion restores compatibility between logic programming and abstract argumentation in dynamic contexts.
• X-SYS: A Reference Architecture for Interactive Explanation Systems — Operationalizing XAI requires treating explainability as an information systems problem with specific user interaction demands.
• We Need Explanation Cards to Connect Explanation Algorithms to the Real World — Explanation Cards bridge the gap between perceived and actual meaning of algorithmic explanations by providing crucial context and interpretation guidance.
• SocialReasoning-Bench: Measuring whether AI agents act in users’ best interests — AI agents often complete tasks but fail to secure optimal outcomes or demonstrate due diligence in social reasoning contexts.
• A Causal Argumentation Method for Explainability of Machine Learning Models — Integrating causal discovery with argumentation frameworks provides "why" explanations for ML model predictions.
• Achieving Gold-Medal-Level Olympiad Reasoning via Simple and Unified Scaling — A compact 30B-A3B model achieves Olympiad-level reasoning through a multi-stage training and inference pipeline.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.MA updates on arXiv.org.