A Survey of Controllable Learning: Methods and Applications in Information Retrieval

2024-07-04

arXiv.org

AI Summary

This paper addresses the lack of controllability and frequent retraining requirements in information retrieval (IR) systems when model responses must dynamically adapt to shifting objectives. We present the first systematic study of Controllable Learning (CL) tailored to IR, formally defining CL and proposing a four-dimensional taxonomy—covering controllable objects, agents, mechanisms, and deployment stages. We introduce, for the first time in IR, a dual-agent (user and platform) and multi-stage (pre-, in-, and post-processing) controllability paradigm. Our framework integrates rule-based modeling, Pareto multi-objective optimization, hypernetwork architectures, and dynamic adaptation techniques across data preprocessing, model training, and re-ranking. Key contributions include: (1) the first comprehensive survey of CL in IR; (2) identification of 12 promising research directions; and (3) identification of emerging themes—including large-model empowerment and trustworthy evaluation—providing both theoretical foundations and practical guidance for building reliable, adaptive IR systems.

Technology Category

• Search and Optimization: Learning to Search
• Machine Learning: Learning Preferences or Rankings
• Intelligent Robots: Behavior Learning & Control

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Controllability has become a crucial aspect of trustworthy machine learning, enabling learners to meet predefined targets and adapt dynamically at test time without requiring retraining as the targets shift. We provide a formal definition of controllable learning (CL), and discuss its applications in information retrieval (IR) where information needs are often complex and dynamic. The survey categorizes CL according to what is controllable (e.g., multiple objectives, user portrait, scenario adaptation), who controls (users or platforms), how control is implemented (e.g., rule-based method, Pareto optimization, hypernetwork and others), and where to implement control (e.g., pre-processing, in-processing, post-processing methods). Then, we identify challenges faced by CL across training, evaluation, task setting, and deployment in online environments. Additionally, we outline promising directions for CL in theoretical analysis, efficient computation, empowering large language models, application scenarios and evaluation frameworks.

Problem

• Controllable Learning
• Machine Learning
• Information Retrieval

Innovation

• Controllable Learning
• Information Retrieval
• Theoretical Framework