KRR-Oxford/DeepOnto

A package for ontology engineering with deep learning and language models.

What it solves

DeepOnto provides a unified Python framework for ontology engineering, bridging the gap between formal knowledge representation (ontologies) and deep learning. It simplifies the process of implementing DL models, constructing resources, and evaluating performance for tasks like ontology matching and subsumption prediction.

How it works

It uses a core Ontology class that integrates with the Java-based OWLAPI via JPype to handle ontology entities and axioms. The framework is organized into several specialized modules:

  • Processing Tools: Includes reasoners for consistency checks, pruners for extracting sub-ontologies, verbalisers to turn logical expressions into text, and projectors/normalisers for embedding preparation.
  • Taxonomy Extraction: Converts ontologies into directed acyclic graphs (DAGs) for graph-based deep learning.
  • Integrated Systems: Bundles specific tools like BERTMap (ontology matching), BERTSubs (subsumption prediction), and OntoLAMA (LLM evaluation for subsumption).

Who it’s for

It is designed for researchers and developers in the fields of semantic web, knowledge engineering, and AI who want to apply deep learning techniques to formal ontologies.

Highlights

  • OWLAPI Integration: Seamlessly connects Python with Java's OWLAPI for robust ontology handling.
  • Comprehensive Toolset: Built-in support for reasoning, pruning, and verbalisation.
  • Pre-integrated Models: Includes BERT-based systems for ontology alignment and subsumption.
  • LLM Probing: Features OntoLAMA for evaluating how language models handle ontology subsumption inference.

Related

  • Project
    deepdoctection/deepdoctection

    A Python library for document understanding that orchestrates layout analysis, OCR, and token classification to extract structured data from PDFs and scans.

  • Project
    Lightning-AI/pytorch-lightning

    A deep learning framework that abstracts away the engineering boilerplate of PyTorch, enabling researchers to pretrain and finetune AI models at scale across various hardware.

  • Project
    jolibrain/deepdetect

    A deep learning runtime and REST server that provides a unified API for training and inference across images, text, and tabular data.

  • Project
    serengil/deepface

    A lightweight Python framework for face recognition and facial attribute analysis that wraps multiple state-of-the-art models for identity verification and demographic prediction.

  • Project
    Fadi002/de4py

    An advanced Python deobfuscator that combines local LLMs with AST cleaning and pattern matching to help malware analysts and reverse engineers recover readable code.