STAR-Q: A web-based platform for annotating qualified knowledge graph statements in texts and tables

The construction and evaluation of knowledge graph (KG) population systems depend on high-quality, entitygrounded annotated corpora. However, producing such corpora remains a laborious task: general-purpose annotation tools do not often support the grounding of annotations to KG identifiers, rarely handle tabular content, and provide little support for corpus-level quality control. In this paper, we present STAR-Q (Span and Table Annotation for RDF with Qualifiers), an open-source, web-based annotation platform designed to facilitate the annotation of both texts and tables with knowledge graph entities. STAR-Q lets users highlight mentions corresponding to subjects, predicates, objects, and ordered qualifier predicate–value pairs directly in documents, and then disambiguate them to entities and properties defined in a reference knowledge graph through a configurable lookup interface targeting any Wikibase instance. Wikidata is supported out of the box, with its large set of QIDs and PIDs. Private Wikibase deployments, such as enterprise knowledge graphs, can also be registered and selected per corpus. Annotations use a unified component schema covering both textual spans and character ranges within table cells; resolved values may be KG resources, corpus-local entities and relations, or typed literals. Beyond annotation, STAR-Q provides corpus management and import adapters for structured JSON/JSONL, archive-based corpus formats, Label Studio exports, and its own versioned full-corpus JSON. Annotations generated by the tool can be exported in RDF 1.2, either in a ”truthy” statement mode or in a ”full” mode preserving component-level text and table evidence using well-known ontologies (DCAT, PROV, NIF, CSVW, Web Annotations). Wikidata-grounded statements can additionally be exported as QuickStatements 3.0 commands for direct submission to Wikidata. We demonstrate STAR-Q through two scenarios: a didactic scenario based on Wookieepedia texts annotated with Wikidata entities curated by the WikiProject Star Wars community effort, and an industrial in-use scenario, where technical telecommunications documents are annotated against an enterprise knowledge graph. The tool is open source at https://github.com/ecladatta/star-q.