Learnijoy
ResearchAI Engineering & DevToolsAI Research

CLARK Framework Enhances Adaptive Reasoning with Knowledge Graphs.

Yousef Khan, Luca Gherardini, Marco Maratea, Joel Arrais, Jose Sousa· July 23, 2026 View original

Summary

CLARK is a new framework that integrates knowledge graphs, symbolic rule mining, and probabilistic reasoning to create adaptive, interpretable, knowledge-driven models. It addresses the limitations of traditional machine learning models in handling uncertain and evolving information by iteratively refining knowledge graphs with learned rules.

Traditional machine learning models often struggle with data distribution shifts and integrating prior knowledge, limiting their effectiveness in dynamic, uncertain environments. A new framework, CLARK (Closed-loop Learning for Adaptive Reasoning over Knowledge Graphs), aims to overcome these challenges by combining several advanced AI techniques. CLARK leverages knowledge graphs, symbolic rule mining, and probabilistic reasoning within the Logic Programs with Markov Logic Networks (LPMLN) formalism. It begins by translating knowledge graph structures into an LPMLN program. The system then iteratively enriches this program by proposing candidate rules from symbolic learners, which are subsequently calibrated using probabilistic weight learning. This closed-loop process allows CLARK to reason under uncertainty and continuously refine the underlying knowledge graph structure. Evaluated on medical datasets, CLARK demonstrated improved classification performance and more generalizable inference, offering a principled approach for building adaptive and interpretable AI systems.

Why it matters

This framework provides a robust method for building AI systems that can adapt to changing information, integrate domain expertise, and offer interpretable reasoning, crucial for high-stakes applications.

How to implement this in your domain

  1. 1Assess existing domain knowledge and data sources for potential representation as knowledge graphs.
  2. 2Investigate LPMLN or similar probabilistic logic programming frameworks for integrating symbolic and statistical reasoning.
  3. 3Pilot CLARK-like methodologies on specific classification tasks where data distribution changes are common.
  4. 4Collaborate with AI researchers to explore the application of adaptive reasoning in complex, evolving datasets.

Who benefits

HealthcareFinanceLegalManufacturing

Key takeaways

  • CLARK integrates knowledge graphs, symbolic rule mining, and probabilistic reasoning for adaptive AI.
  • It addresses limitations of traditional ML in handling uncertain and evolving data.
  • The framework iteratively refines knowledge graphs with learned and calibrated rules.
  • CLARK improves classification performance and generalizable inference, especially in medical domains.

Original post by Yousef Khan, Luca Gherardini, Marco Maratea, Joel Arrais, Jose Sousa

"arXiv:2607.19996v1 Announce Type: new Abstract: Machine Learning models are widely used for automating classification tasks by extracting statistical patterns from data. However, their performance deteriorates if the data distribution changes, making them ill-suited to handle unc…"

View on X

Originally posted by Yousef Khan, Luca Gherardini, Marco Maratea, Joel Arrais, Jose Sousa on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses

More in AI Engineering & DevTools

AI ResearchAI Engineering & DevTools

New Adaptive Filter Improves Time-Series Prediction with Input Noise

Researchers developed the RFFBCGA algorithm, a new nonlinear adaptive filter that effectively mitigates both input and output noise in time-series prediction. This method maintains a fixed network structure while enhancing robustness across various noise scenarios.

Yi Peng, Haiquan Zhao, Jinhui HuJul 23, 2026
AI ResearchAI Engineering & DevTools

New Algorithm Learns Local Causal Structures with Latent Variables

Researchers propose LoCaLS, a new algorithm for learning local causal structures around a target variable from observational data, even when latent variables and selection bias are present. LoCaLS achieves high accuracy with significantly less computational effort than global causal discovery methods.

Zheng Li, Hao Zhang, Ruxin Wang, Ruichu Cai, Kun Zhang, Feng XieJul 23, 2026
AI Engineering & DevToolsAI Research

New Framework Evaluates AI Robustness with Minimum-Norm Attacks

Researchers introduce a unified framework for evaluating adversarial robustness using a comprehensive pool of minimum-norm attacks and robustness-perturbation curves across multiple norms. This approach addresses limitations of fixed-epsilon evaluations, providing a more stable and controllable assessment of AI model defenses.

Luca Scionis, Luca Melis, Maura Pintor, Fabio Brau, Ambra Demontis, Giorgio Fumera, Fabio Roli, Battista BiggioJul 23, 2026