Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Construct, Align, and Reason: Large Ontology Models for Enterprise Knowledge Management

Yao Zhang, Hongyin Zhu

TL;DR

The paper addresses the challenge of enterprise knowledge management across heterogeneous structured and unstructured data, which hampers deep semantic reasoning. It introduces the Large Ontology Model (LOM), a construct--align--reason framework that unifies ontology construction, text-ontology alignment, and instruction-aligned reasoning, trained via ontology instruction tuning, text-ontology grounding, and curriculum-based multi-task tuning, backed by a CoT-enhanced graph-reasoning dataset. The approach delivers a dual-layer ontology from structured and unstructured sources, a cross-source alignment mechanism, and a 4B-parameter model achieving 89.47% accuracy on 19 graph-reasoning tasks, outperforming baselines. This work demonstrates robust, structure-aware reasoning over enterprise data, enabling scalable, multi-hop inference for practical knowledge-management applications.

Abstract

Enterprise-scale knowledge management faces significant challenges in integrating multi-source heterogeneous data and enabling effective semantic reasoning. Traditional knowledge graphs often struggle with implicit relationship discovery and lack sufficient semantic understanding for complex question answering. To address these limitations, we introduce a unified construct--align--reason framework, the large ontology model (LOM). We first build a dual-layer enterprise ontology from structured databases and unstructured text, subsequently fusing these sources into a comprehensive enterprise ontology. To enable instruction-aligned reasoning, we propose a unified three-stage training pipeline: ontology instruction fine-tuning to improve structural understanding; text-ontology grounding to strengthen node semantic encoding; and multi-task instruction tuning on ontology-language pairs with curriculum learning to enhance semantic reasoning and generation. We also construct comprehensive training and evaluation datasets covering diverse ontology reasoning tasks. On this benchmark, our 4B-parameter LOM achieves 89.47% accuracy and outperforms DeepSeek-V3.2 on complex graph reasoning, indicating effective fusion of ontology structure and language.

Construct, Align, and Reason: Large Ontology Models for Enterprise Knowledge Management

TL;DR

The paper addresses the challenge of enterprise knowledge management across heterogeneous structured and unstructured data, which hampers deep semantic reasoning. It introduces the Large Ontology Model (LOM), a construct--align--reason framework that unifies ontology construction, text-ontology alignment, and instruction-aligned reasoning, trained via ontology instruction tuning, text-ontology grounding, and curriculum-based multi-task tuning, backed by a CoT-enhanced graph-reasoning dataset. The approach delivers a dual-layer ontology from structured and unstructured sources, a cross-source alignment mechanism, and a 4B-parameter model achieving 89.47% accuracy on 19 graph-reasoning tasks, outperforming baselines. This work demonstrates robust, structure-aware reasoning over enterprise data, enabling scalable, multi-hop inference for practical knowledge-management applications.

Abstract

Enterprise-scale knowledge management faces significant challenges in integrating multi-source heterogeneous data and enabling effective semantic reasoning. Traditional knowledge graphs often struggle with implicit relationship discovery and lack sufficient semantic understanding for complex question answering. To address these limitations, we introduce a unified construct--align--reason framework, the large ontology model (LOM). We first build a dual-layer enterprise ontology from structured databases and unstructured text, subsequently fusing these sources into a comprehensive enterprise ontology. To enable instruction-aligned reasoning, we propose a unified three-stage training pipeline: ontology instruction fine-tuning to improve structural understanding; text-ontology grounding to strengthen node semantic encoding; and multi-task instruction tuning on ontology-language pairs with curriculum learning to enhance semantic reasoning and generation. We also construct comprehensive training and evaluation datasets covering diverse ontology reasoning tasks. On this benchmark, our 4B-parameter LOM achieves 89.47% accuracy and outperforms DeepSeek-V3.2 on complex graph reasoning, indicating effective fusion of ontology structure and language.
Paper Structure (20 sections, 8 equations, 3 figures, 1 table)

This paper contains 20 sections, 8 equations, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Ontology Construction from Heterogeneous Data
  4. Ontology Reasoning
  5. Approach
  6. Large Ontology Model
  7. Model Architecture
  8. Training Method
  9. Dataset Construction
  10. Ontology Construction
  11. Ontology Construction from Structured Data
  12. Ontology Construction from Unstructured Data
  13. Experiments
  14. Dataset
  15. Hyper-parameters
  16. ...and 5 more sections

Figures (3)

  • Figure 1: Model Architecture Overview
  • Figure 2: Model Performance Ranking
  • Figure 3: Model Performance Heatmap