ontology

Proposes OPI, an ontology-guided framework for multi-hop knowledge graph question answering that leverages a relation-centric ontology graph for bidirectional retrieval and iterative refinement, achieving state-of-the-art results on multiple benchmarks.

This paper presents JD Oxygen AI Item Center (Oxygen AIIC), an industrial-scale platform leveraging LLMs/VLMs for item knowledge production and service, achieving high precision and recall, and delivering measurable gains in search, recommendation, and operations on JD.com.

TOTEN is a knowledge-based ontological tokenization framework that replaces statistical tokenization with declarative classification grounded in a formal ontology of engineering entities, achieving high ontological atomicity and numerical reconstruction for physical quantities and technical notation in Brazilian Portuguese.

This paper proposes AgenticRei, a framework for runtime governance of LLM-driven agentic AI systems using deontic policies expressed in OWL, enabling obligations, dispensations, and conflict resolution beyond traditional policy engines.

This paper introduces a structured ontology for untranslatability in machine translation, along with a taxonomy of compensation strategies and a multilingual dataset. Human preference studies show translator quality depends on the strategy used, with a preference for explanatory translations.

The author describes a problem where an LLM agent's memory graph gets corrupted by incorrect edges, and proposes using a declared ontology to validate writes and traversals. A test on 120 deliberately broken traversals caught all errors.

Researchers present an ontology-grounded framework for pre-deployment verification of enterprise AI agents, combining an Agent Operational Envelope, automated scenario generation, and machine-verifiable Trust Certificates with graduated deployment verdicts. A pilot across four regulated industries generated 1,800 scenarios and showed ontology-grounded generation significantly outperformed persona-based baselines on regulatory coverage.

The Violation Situation Pattern (VSP) reifies compliance violations as persistent graph nodes with lifecycle states and audit history, enabling durable, queryable violation records that can evolve detection logic without invalidating accumulated history.

This paper proposes a neuro-symbolic framework for constructing ontology-grounded knowledge graphs from text by deferring consistency corrections to a post-extraction stage, reducing token usage while improving KG consistency and maintaining QA performance.

Explains how to fix agent memory by defining an ontology using Pydantic schemas, enabling structured extraction into knowledge graphs for multi-hop reasoning, with an open-source solution (Zep).

SciCustom is a framework for constructing custom scientific benchmarks from large-scale data, enabling fine-grained evaluation of LLMs' scientific capabilities without expert annotation. It uses ontology-grounded knowledge units and voting-based consensus to select relevant benchmarks, demonstrated in chemistry and healthcare.

This paper proposes a symbolic framework that converts redacted police narratives into evidence-linked facts using ontology, semantic parsing (AMR), and reasoning, enabling structured querying of incident details that are typically only available in free text.

HEAR is an enterprise agentic reasoner using a Stratified Hypergraph Ontology to perform multi-hop reasoning over heterogeneous business systems, achieving up to 94.7% accuracy on supply-chain tasks.

This paper introduces SAGE, a hierarchical LLM-based framework for evaluating literary quality through ontology-grounded interpretive dimensions. It demonstrates high reliability and inter-rater agreement in assessing cultural, emotional, and philosophical aspects of narratives, highlighting gaps between human-authored and LLM-generated works.

The article analyzes how Palantir's AIP platform serves as an example of an end-to-end Agent architecture, emphasizing the importance of ontology as the core enterprise knowledge for building an evolvable intelligence system.