Knowledge-driven Augmentation and Retrieval for Integrative Temporal Adaptation

2026-04-23

AI Summary

This work addresses the challenge of jointly modeling semantic distribution shifts and domain knowledge evolution in temporally evolving scenarios, a setting where existing methods struggle to capture complex temporal dynamics. To this end, we propose KARITA, a novel framework that, for the first time, deeply integrates structured domain knowledge—such as MeSH ontologies—into the temporal adaptation process. By leveraging knowledge graph embeddings and multi-source knowledge fusion, KARITA unifies the modeling of semantic and knowledge co-evolution over time. The framework incorporates a temporally informed, selective retrieval-augmented learning mechanism that dynamically enhances cross-temporal generalization by accounting for both predictive uncertainty and feature drift patterns. Extensive experiments on clinical, legal, and scientific text classification benchmarks demonstrate that KARITA significantly outperforms state-of-the-art baselines across multiple temporal datasets, underscoring the critical role and superiority of knowledge-driven strategies in temporal domain adaptation.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Data Mining & Knowledge Management: Mining of Spatial, Temporal or Spatio-Temporal Data
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

Time introduces fundamental challenges in model development and deployment: models are usually trained on historical data while deployed on future data where semantic distributions and domain knowledge may evolve. Unfortunately, existing studies either overlook temporal shifts or hardly capture rich shifting patterns of both semantic and knowledge. We develop Knowledge-driven Augmentation and Retrieval for Integrative Temporal Adaptation (KARITA) to capture diverse temporal shifts (e.g., uncertainty and feature shift), construct and integrate rich knowledge sources (e.g., medical ontology like MeSH), and leverage shifting insights for selecting-retrieval augmented learning. We evaluate KARITA on classification tasks across multiple domains, clinical, legal, and scientific corpora, demonstrating consistent improvements across multiple domains with temporal adaptation. Our results show that knowledge integration can be more critical and effective in temporal augmentation and learning.

Problem

• temporal shift
• semantic distribution
• domain knowledge
• temporal adaptation
• knowledge integration

Innovation

• temporal adaptation
• knowledge integration
• retrieval-augmented learning
• domain shift
• knowledge-driven augmentation