Descriptive History Representations: Learning Representations by Answering Questions

2025-06-02

AI Summary

In partially observable environments, long interaction histories pose challenges for effective compression while preserving optimality in decision-making. To address this, we propose Descriptive Historical Representations (DHRs)—the first sufficient statistical representation formally defined by the principle of *answerability*. Methodologically, we design a multi-agent framework that jointly trains representation learning, policy optimization, and question generation, integrating question-driven representation learning, text generation, and user modeling. Experiments on public movie and shopping datasets demonstrate that the generated textual user profiles significantly improve preference behavior prediction accuracy (+8.2%–12.7%) and exhibit strong interpretability and task-oriented utility. Our core contribution is establishing answerability as a novel paradigm for representation sufficiency, enabling joint optimization of historical compression, decision control, and human-interpretable interaction.

Technology Category

• Knowledge Representation and Reasoning: Preferences
• Natural Language Processing: Question Answering
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Effective decision making in partially observable environments requires compressing long interaction histories into informative representations. We introduce Descriptive History Representations (DHRs): sufficient statistics characterized by their capacity to answer relevant questions about past interactions and potential future outcomes. DHRs focus on capturing the information necessary to address task-relevant queries, providing a structured way to summarize a history for optimal control. We propose a multi-agent learning framework, involving representation, decision, and question-asking components, optimized using a joint objective that balances reward maximization with the representation's ability to answer informative questions. This yields representations that capture the salient historical details and predictive structures needed for effective decision making. We validate our approach on user modeling tasks with public movie and shopping datasets, generating interpretable textual user profiles which serve as sufficient statistics for predicting preference-driven behavior of users.

Problem

• Compress long interaction histories into informative representations
• Answer relevant questions about past and future outcomes
• Generate interpretable user profiles for behavior prediction

Innovation

• Descriptive History Representations for decision making
• Multi-agent learning framework with joint optimization
• Interpretable textual user profiles for behavior prediction