Treffer: ESEA (East-and-Southeast-Asian) Traditional Music Knowledge Base and Its Ontology-Subgraph-Driven NLQ2SPARQL Intelligent Question-Answering System.

We introduce an East-and-Southeast Asian Traditional Music Knowledge Base (ESEA_TM), which, along with the LinkedMusic project, is based on linked data and aims to enhance access across various music databases via AI. ESEA_TM was initially built upon ontology engineering, focusing on cataloging and classification of controlled vocabularies and semantic queries, for example, featuring music classification by music type (music species), and highlighting the use of LLMs to extract traditional instrument entries. We demonstrate the general ontology structure and semantic query examples for ESEA_TM using SPARQL, and especially queries based on knowledge inference. Moreover, in terms of converting Natural Language Questions to SPARQL (NLQ2SPARQL), we provided OWL snippets to LLMs for SPARQL generation. The issue is that, for an oversized ontology, LLMs struggle to pinpoint the corresponding OWL snippet—an ontology subgraph—for correct generation. To address this, we introduce a "subgraph extraction from ontology" approach. The workflow (in Python script) is: (1) Specific Ontology Editing, (2) Ontology Segmentation, (3) Entity Extraction from the Ontology Segments, (4) Subgraph Assembly, (5) SPARQL Generation and Verification Based on Subgraph, (6) Retrieval Augmented Generation (RAG) and Recommendation Based on SPARQL Syntax, (7) Retrieval Recommendation Based on the Neighborhood in Ontology Subgraph. Finally, with cases testing summary, we assess the feasibility, discuss improvement direction, and reflect upon a comparison among different orientations for NLQ2SPARQL. [ABSTRACT FROM AUTHOR]

Copyright of Cataloging & Classification Quarterly is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)