Merge pull request #71 from sciknoworg/main

rebase dev with main

Author: HamedBabaei

CHANGELOG.md

@@ -1,5 +1,10 @@
 ## Changelog
 
+### V1.5.2 Changelog (October 12, 2025)
+- Fix to `bitsandbytes` install for MacOS -- it will only be installed for linux.
+- Add KGE retriever
+- Update KGE documentation
+
 ### V1.5.1 Changelog (September 7, 2025)
 - Update dependencies.
 - Make versioning standardized.

CITATION.cff

@@ -17,5 +17,5 @@ keywords:
   - "Alignment"
   - "Python Library"
 license: "Apache-2.0"
-version: "1.5.1"
-date-released: "2025-07-29"
+version: "1.5.2"
+date-released: "2025-10-12"

docs/source/aligner/rag.rst

@@ -1,24 +1,27 @@
 Retrieval-Augmented Generation
 ================================
 
-LLMs4OM
-----------------------------------
-**LLMs4OM: Matching Ontologies with Large Language Models**
 
 .. sidebar:: **Reference:**
 
     `LLMs4OM: Matching Ontologies with Large Language Models <https://link.springer.com/chapter/10.1007/978-3-031-78952-6_3>`_
 
-The retrieval augmented generation (RAG) module at OntoAligner is driven by a ``LLMs4OM`` framework, a novel approach for effective ontology  alignment using LLMs. This framework utilizes two modules for retrieval and matching, respectively, enhanced by zero-shot prompting across three ontology representations: concept, concept-parent, and concept-children.  The ``LLMs4OM`` framework, can match and even surpass the performance of traditional OM systems, particularly in complex matching scenarios. The following diagram represent the ``LLMs4OM`` framework.
+	.. raw:: html
+
+		<iframe src="https://videolectures.net/embed/videos/eswc2024_babaei_giglou_language_models?part=1" width="100%" frameborder="0" allowfullscreen style="aspect-ratio:16/9"></iframe>
+
+LLMs4OM
+----------------------------------
+**LLMs4OM: Matching Ontologies with Large Language Models**
+
+The retrieval augmented generation (RAG) module at OntoAligner is driven by a ``LLMs4OM`` framework, a novel approach for effective ontology  alignment using LLMs. This framework utilizes two modules for retrieval and matching, respectively, enhanced by zero-shot prompting across three ontology representations: concept, concept-parent, and concept-children.  The ``LLMs4OM`` framework, can match and even surpass the performance of traditional OM systems, particularly in complex matching scenarios. The ``LLMs4OM`` framework (as shown in the following diagram) offers a RAG approach within LLMs for OM. LLMs4OM uses :math:`O_{source}` as query :math:`Q(O_{source})` to retrieve possible matches for for any :math:`C_s \in C_{source}` from :math:`C_{target} \in O_{target}`. Where, :math:`C_{target}` is stored in the knowledge base :math:`KB(O_{target})`. Later, :math:`C_{s}` and obtained :math:`C_t \in C_{target}` are used to query the LLM to check whether the :math:`(C_s, C_t)` pair is a match. As shown in above diagram, the framework comprises four main steps: 1) Concept representation, 2) Retriever model, 3) LLM, and 4) Post-processing. But within the OntoAligner we we adapted the workflow into a parser, encoder, alignment, post-processing, evaluate, and export steps.
 
 .. raw:: html
 
     <div align="center">
      <img src="https://raw.githubusercontent.com/sciknoworg/OntoAligner/refs/heads/dev/docs/source/img/LLMs4OM.jpg" width="80%"/>
     </div>
 
-The ``LLMs4OM`` framework offers a RAG approach within LLMs for OM. LLMs4OM uses :math:`O_{source}` as query :math:`Q(O_{source})` to retrieve possible matches for for any :math:`C_s \in C_{source}` from :math:`C_{target} \in O_{target}`. Where, :math:`C_{target}` is stored in the knowledge base :math:`KB(O_{target})`. Later, :math:`C_{s}` and obtained :math:`C_t \in C_{target}` are used to query the LLM to check whether the :math:`(C_s, C_t)` pair is a match. As shown in above diagram, the framework comprises four main steps: 1) Concept representation, 2) Retriever model, 3) LLM, and 4) Post-processing. But within the OntoAligner we we adapted the workflow into a parser, encoder, alignment, post-processing, evaluate, and export steps.
-
 
 Usage
 ----------------

docs/source/index.rst

@@ -17,6 +17,11 @@ OntoAligner was created by `Scientific Knowledge Organization (SciKnowOrg group)
       <strong>The vision is to create a unified hub that brings together a wide range of ontology alignment models, making integration seamless for researchers and practitioners.</strong>
     </div>
 
+**Watch the OntoAligner presentation at EWC-2025.**
+
+.. raw:: html
+
+	<iframe src="https://videolectures.net/embed/videos/eswc2025_bernardin_babaei_giglu?part=1" width="100%" frameborder="0" allowfullscreen style="aspect-ratio:16/9"></iframe>
 
 Citing
 =========
@@ -47,6 +52,19 @@ or our related work `LLMs4OM: Matching Ontologies with Large Language Models <ht
       organization={Springer}
     }
 
+or if you are using Knowledge Graph Embeddings refer to `OntoAligner Meets Knowledge Graph Embedding Aligners <https://arxiv.org/abs/2509.26417>`_:
+
+ .. code-block:: bibtex
+
+  @article{babaei2025ontoaligner,
+	  title={OntoAligner Meets Knowledge Graph Embedding Aligners},
+	  author={Babaei Giglou, Hamed and D'Souza, Jennifer and Auer, S{\"o}ren and Sanaei, Mahsa},
+	  journal={arXiv e-prints},
+	  pages={arXiv--2509},
+	  year={2025}
+	}
+
+
 
 .. toctree::
    :maxdepth: 1

ontoaligner/VERSION

@@ -1 +1 @@
-1.5.1
+1.5.2