Newest papers

When More Metadata Misleads: Probing Field-Level Grounding in LLM-Generated Museum Descriptions

Clemencia Siro, Laura Hollink A · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.31193 · cited by 0 works (OpenAlex) · open-access PDF

German

Metadatenbedingte Generation wird oft als eine Daten-zu-Text-Aufgabe eingeführt, in der zusätzliche Eingabefelder die Grundlegung verbessern sollten. Museumbeschreibung Generation fordert diese Annahme heraus, weil Katalogaufzeichnungen heterogen, teilweise visuell und ungleichmäßig in geografischen Regionen dokumentiert sind. Wir untersuchen, wie Metadatenzusammensetzung die Generation von Artefaktbeschreibungen durch einen zweistufigen Rahmen formt. Phase 1 analysiert natürliche Metadatenvariationen in einer großen Museumssammlung, wo Feldzählung, Feldtyp, Bildverfügbarkeit und Platzkovariation stattfinden. Phase 2 führt kontrollierte Feldsonden ein, die Objektidentität festhalten, während sich die Metadatenmenge, Feldtyp und Bildzugang untersche

Machine translation (Meta M2M100); the English original follows.

Abstract

Metadata-conditioned generation is often framed as a data-to-text task in which additional input fields should improve grounding. Museum description generation challenges this assumption because catalogue records are heterogeneous, partially visual, and unevenly documented across geographic regions. We study how metadata composition shapes the generation of artefact descriptions through a two-phase framework. Phase 1 analyses natural metadata variation in a large museum collection, where field count, field type, image availability, and Place co-vary. Phase 2 introduces controlled field probes that hold object identity fixed while varying metadata quantity, field type, and image access. Our results show that metadata quantity alone does not explain generation behaviour. Field type determines whether models ground, omit, or infer information, while images can both support visual grounding and increase unsupported interpretation. More broadly, our findings show that metadata-conditioned data-to-text evaluation should assess field- and claim-level support, rather than relying only on aggregate input coverage or reference similarity.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Metadata, Computer science, Field (mathematics), Information retrieval, Task (project management), Object (grammar)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange