Add the BERTSub Module

The major update of this release is the deployment of the BERTSubs system, available at deeponto.subs.bertsubs. Besides, we move the pruning functionality out of Ontology to form a separate module for future development. See more details at changelog.

More OWLAPI Features

This update is mainly about the base class deeponto.onto.Ontology where more OWLAPI features are encapsulated into handy functions (see changelog).

Officially release the Ontology Verbaliser and OntoLAMA modules

Officially release the ontology verbaliser module at deeponto.onto.verbalisation.

Fully deploy deeponto.lama referring to the paper: Language Model Analysis for Ontology Subsumption Inference. The current version of the paper has not covered the information about the new verbaliser, it will be updated soon.

Fix Minor Bugs for Reasoning

• Fix the minor bugs in the onto.reasoner.instances_of() method.
• Introduce normalise_identifiers argument in onto.build_annotation_index() such that it becomes an optional pre-processing step.

Add JVM Memory Prompt

Importing ontology now requires users to input the maximum JVM memory allocation (defaults to 8g).

Add the Ontology Verbalisation Module; Fix PyPI layout.

An ontology verbalisation tool (see everything in deeponto.onto.verbalisation) is introduced in this release, along with a syntax parser and a datastructure called RangeNode. This verbalisation tool supports a limited set of class expression patterns, but is suitable for common complex class expressions without too many nested levels.

The layout indexed in PyPI was faulty in the previous release and is now fixed.

Brand New DeepOnto

$\textsf{DeepOnto}$ is completely rebuilt on the OWLAPI, and becoming an officially released package. As the first new-born version, it supports the basic ontology features, ontology alignment systems including BERTMap and BERTMapLt, and reproducing the OM resource construction as proposed in Bio-ML.

A known issue of this release is the faulty PyPI layout.