Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Three-stage Neuro-symbolic Recommendation Pipeline for Cultural Heritage Knowledge Graphs

Krzysztof Kutt, Elżbieta Sroka, Oleksandra Ishchuk, Luiz do Valle Miranda

TL;DR

This paper presents a complete methodology for implementing a hybrid recommendation pipeline integrating knowledge-graph embeddings, approximate nearest-neighbour search, and SPARQL-driven semantic filtering and presents the final three-stage neuro-symbolic recommender.

Abstract

The growing volume of digital cultural heritage resources highlights the need for advanced recommendation methods capable of interpreting semantic relationships between heterogeneous data entities. This paper presents a complete methodology for implementing a hybrid recommendation pipeline integrating knowledge-graph embeddings, approximate nearest-neighbour search, and SPARQL-driven semantic filtering. The work is evaluated on the JUHMP (Jagiellonian University Heritage Metadata Portal) knowledge graph developed within the CHExRISH project, which at the time of experimentation contained ${\approx}3.2$M RDF triples describing people, events, objects, and historical relations affiliated with the Jagiellonian University (Kraków, PL). We evaluate four embedding families (TransE, ComplEx, ConvE, CompGCN) and perform hyperparameter selection for ComplEx and HNSW. Then, we present and evaluate the final three-stage neuro-symbolic recommender. Despite sparse and heterogeneous metadata, the approach produces useful and explainable recommendations, which were also proven with expert evaluation.

A Three-stage Neuro-symbolic Recommendation Pipeline for Cultural Heritage Knowledge Graphs

TL;DR

This paper presents a complete methodology for implementing a hybrid recommendation pipeline integrating knowledge-graph embeddings, approximate nearest-neighbour search, and SPARQL-driven semantic filtering and presents the final three-stage neuro-symbolic recommender.

Abstract

The growing volume of digital cultural heritage resources highlights the need for advanced recommendation methods capable of interpreting semantic relationships between heterogeneous data entities. This paper presents a complete methodology for implementing a hybrid recommendation pipeline integrating knowledge-graph embeddings, approximate nearest-neighbour search, and SPARQL-driven semantic filtering. The work is evaluated on the JUHMP (Jagiellonian University Heritage Metadata Portal) knowledge graph developed within the CHExRISH project, which at the time of experimentation contained M RDF triples describing people, events, objects, and historical relations affiliated with the Jagiellonian University (Kraków, PL). We evaluate four embedding families (TransE, ComplEx, ConvE, CompGCN) and perform hyperparameter selection for ComplEx and HNSW. Then, we present and evaluate the final three-stage neuro-symbolic recommender. Despite sparse and heterogeneous metadata, the approach produces useful and explainable recommendations, which were also proven with expert evaluation.
Paper Structure (18 sections, 1 figure, 4 tables)

This paper contains 18 sections, 1 figure, 4 tables.

Table of Contents

  1. Introduction
  2. Related works
  3. Cultural heritage knowledge graphs in practice
  4. Recommendation systems not only for cultural heritage data
  5. Methods
  6. The JUHMP knowledge graph and CHExRISH context
  7. Experimental setups
  8. Three-stage recommendation pipeline
  9. Embedding model selection and hyperparameters
  10. HNSW parameters comparison
  11. SPARQL-driven semantic filtering
  12. Evaluation approach
  13. Results and Discussion
  14. Quantitative evaluation
  15. Expert evaluation
  16. ...and 3 more sections

Figures (1)

  • Figure 1: Expert questionnaire. Question about target Benedykt from Koźmin Wielkopolski with the top 10 results generated by the pipeline. The experts were asked to assess the correctness of all 10 results with $[2, 1, 0, -1]$ scale described in the text.