Webinar: ImageSnippets: Experiences with a Linked Data Image Annotation System (SIG-CMR)

Description

In metadata schemas, descriptive information about images is often either condensed into keywords and subject classifications or stored in description fields that, despite being conceptually rich and contextually relevant, are typically formatted only as plain text. In this webinar, Margaret Warren will discuss her decades-long research into modeling image descriptions as structured data using the architectural affordances of RDF. Through her experiences with ImageSnippets, a native linked data annotation system, she will discuss topics such as: precision in triple based descriptions, best practices for entity resolution across multiple datasets, and AI/LLM augmentation. She will also show the variety of ways in which this data, once stored in image based knowledge graphs and linked to external sources, can be used for ontology engineering, search and retrieval, publishing, research and even as a means of analyzing the explainability across inferred paths through graph structures (such as those found in Wikidata.)