Overview

Integrating Knowledge Engineering (KE) with Machine Learning (ML) offers a promising approach to building trustworthy AI systems. By combining the strengths of data-driven learning with structured knowledge—such as electronic health records in healthcare, scientific axioms, or legal guidelines—AI systems gain the ability to perform commonsense reasoning, enhancing their reliability and making them more knowledge-aware. Although using knowledge representation and reasoning methods can sometimes limit scalability, their ability to provide verifiable, human-understandable explanations makes them especially valuable in mission-critical applications.

The workshop hosted by HHAI 2026 seeks to bridge the gap between KE and ML by exploring the synergies between these fields. A key focus is on developing hybrid human-AI systems that utilize multimodal approaches, incorporating various forms of data including text, speech, images, and video. This collaborative forum will bring together researchers and practitioners from academia and industry to discuss cutting-edge research and innovative strategies for integrating KE and ML. Ultimately, the goal is to advance the development of AI systems that are not only robust and efficient but also transparent and human-centric, addressing both the challenges and benefits of merging knowledge representation and reasoning with data-driven techniques.

Check out our previous year program and proceedings here HHAI-KEML 2025 Workshop

Important Dates

• Paper Submission: May 16, 2026 (23:59 AoE)
• Acceptance Notification: May 23, 2026
• Camera-ready Version: May 29, 2026
• Workshop: July 7, 2026
• Conference: July 8–10, 2026

Organizing & Program Committees

Organizing Committee

• Shreya Banerjee, University of New Orleans
• Atriya Sen, Oklahoma State University
• Yuan Lin, Kristiania University College

Program Committee

• Shreya Banerjee, University of New Orleans
• Atriya Sen, Oklahoma State University
• Yuan Lin, Kristiania University College
• Anthony Marchiafava, Oklahoma State University
• Henry Fordjour Ansah, University of New Orleans
• Bithi Banik, Kristiania University College

Topics of Interest

• Neuro-Symbolic Knowledge Representation
• KRR and Commonsense Reasoning in ML
• Artificial General Intelligence
• KE with Generative AI and Large Language Models
• KE in Multi-Agent AI Systems
• KRR in Hybrid AI Systems for Health and Other Application Domains
• Human-Centric Multimodal AI
• Human-AI Interaction and Human-in-the-loop
• Knowledge Representation in World Model
• Federated and Privacy-Preserving Learning in Hybrid AI
• Evaluation, Interpretability, and Trustworthy AI

Paper Submission & Review Process

Authors should submit papers electronically in PDF format to Easy Chair. Please use CEURART style formatting to write single column papers to be published with CEUR-WS.

Paper Types:

• Regular Paper: 10-12 pages (excluding references)
• Short Paper: 5–9 pages (excluding references)
• Abstract/Extended Abstract: Title, author, and abstract only
• Poster Paper: A title, author and a very short text (less than 5 “standard” CEUR-WS pages). Poster papers will mostly be handled like an abstract.

All submissions will be peer-reviewed through a double-blind process. Papers should be anonymized and written in English. Submissions of regular and short papers should be original work without substantial overlap with pre-published papers. For further details, please see the submission guidelines.

Accepted submissions shall be submitted to CEUR-WS.org for online publication in CEUR Workshop Proceedings (Scopus indexed). Contributions will be presented either as oral presentations (lightning talks) or posters.
Note: Though in-person participation is highly encouraged, in case of travel-related difficulties, remote attendance and presentations are permitted.

Speakers

Keynote Speaker: Dr. Selmer Bringsjord

Talk: Logic’s Rescue of Numerical Machine Learning

Dr. Selmer Bringsjord (http://kryten.mm.rpi.edu/selmerbringsjord.html) is a professor of Computer Science and Cognitive Science at Rensselaer Polytechnic Institute. He is the director of Rensselaer AI & Reasoning (RAIR) Lab (https://rair.cogsci.rpi.edu/), now going strong for over two decades. Dr. Bringsjord specializes in the logico-mathematical and philosophical foundations of artificial intelligence (AI) and cognitive science (CogSci), in collaboratively building AI systems/robots on the basis (primarily) of computational logic, and in the logic-based and theorem-guided modeling and simulation of rational, human-level-and-above cognition. Work in these areas has been expressed e.g. in over 300 refereed papers/chapters, and books, pursued as an investigator in sponsored-research awards of over $28M, and communicated/debated in person on 6 continents and 33 countries. Though Bringsjord spends considerable engineering time in pursuit of ever-smarter Turing-level computing machines for his much-appreciated sponsors, he claims that “armchair” reasoning time has enabled him to deduce that the human mind will forever be superior to such machines. Bringsjord received the bachelor’s degree from the University of Pennsylvania, where he was heavily influenced by James Ross, and the PhD from Brown University, where he studied under Roderick Chisholm (as did Ross himself). Bringsjord is not unhappy about the apparent fact that he is through Chisholm an intellectual descendant of Leibniz, many of whose views to a high degree align with his own, and whose interest in a rather wide range of intellectual matters matches his own trans-disciplinary modus operandi. Bringsjord claims to have discovered what Leibniz sought throughout his life: the art of infallibility (le art d’infaillibilit ́e = “The Art”), which is composed of an underlying language (the characteristica universalis) and an ensemble of computational reasoning systems (calculus ratiocinator), and can be used to calmly and enjoyably settle by rational adjudication all manner of dispute. Bringsjord has long been on faculty at America’s oldest technological university: Rensselaer Polytechnic Institute (RPI) in Troy NY; where he currently holds appointments in the Department of Cognitive Science, the Department of Computer Science, and the Lally School of Management, and where as a Full Professor he teaches AI, formal logic, formal human and machine reasoning and decision-making (and applications thereof, e.g. in nuclear strategy and micro-economics), and philosophy of AI and CogSci. Funding for Bringsjord’s r&d has come from the Luce Foundation, the National Science Foundation, the Templeton Foundation, AT&T, IBM, Apple, AFRL, ARDA/DTO/IARPA, ONR, DARPA, AFOSR, France’s ANR, and other sponsors. Bringsjord has consulted to and advised many companies in the general realm of intelligent systems, and continues to do so.

Abstract

From a formal perspective we haven’t learned why deep neural networks sometimes generate patently illogical content (e.g. hallucinated references) when the user of such a system wants nothing of the sort. But we’ve recently learned (from e.g. Anthropic’s source-code leak) that because of this ignorance and content, some AI companies have proprietarily, prudently, and secretly brought to bear a makeshift mosaic in hopes of patching the problem. This mosaic is composed of many parts that are logic-based, hence unrelated to deep learning, and for that matter to any kind of numerical ML. But a makeshift mosaic is obviously not optimal. What ought to be done in principled fashion to rationalize a foundation-model AI? I consider and assess some answers to this question; they all of course have in common that they rely upon logic-based AI (and its sub-parts, e.g. knowledge graphs & ontologies) to come to the rescue. I end by explaining that ultimately the only way available to understand foundation-model AIs is to formalize them —— and formalization of anything is by definition to reduce that thing to logic.

Tentative Program Schedule