Figure 10
The diagram presents a left to right workflow with labeled shapes and directed arrows. At the far left bottom, an oval labeled “Document” has the word “START” written below it. An upward arrow from “Document” points to a diamond labeled “is a C H item discussed?”. From this diamond, an upward arrow labeled “NO ITEM” points to an oval labeled “Document is skipped”. From the diamond labeled “is a C H item discussed?”, a rightward arrow labeled “IF ITEM” points to a parallelogram labeled “Extract Metadata of Item”. A dotted rightward arrow labeled “INJECT METADATA” extends from “Extract Metadata of Item” toward a diamond labeled “is the entity a expressing an opinion on the C H item?”. From the oval labeled “Document”, a rightward arrow labeled “GLINER” points to a parallelogram labeled “Identify all entities with GLINER plus Wikidata E L”. Beneath this box appears the text “WIKIDATA” with its logo. From “Identify all entities with GLINER Wikidata E L”, a rightward arrow labeled “Entity plus Wikidata Context” points to the diamond labeled “is the entity a expressing an opinion on the C H item?”. From the diamond labeled “is the entity a expressing an opinion on the C H item?”, three dotted rightward arrows branch to three parallelograms labeled “Extract statement provenance”, “Extract evidences”, and “Extract hypotheses”. Solid rightward arrows from these three boxes converge into a trapezoid labeled “Merge K G”. From “Merge K G”, a rightward arrow leads to three vertically arranged parallelograms grouped by dotted bracket lines. The top box is labeled “Use skos to reconcile strings with S E B I individuals”. The middle box is labeled “Use Wikidata to enhance K G”. The bottom box is labeled “Map to R D F using a queue approach”. A rightward arrow from this grouped section points to an oval labeled “S E B I dash K G”. To the right of this oval, a small node connects to several circular nodes forming a graph structure. All arrows clearly indicate direction from left to right or upward as labeled, and all processing steps and decision paths are connected by arrowheads showing flow direction. Additional visual elements appear in the diagram. Above the “Extract Metadata of Item” region, there is a small illustration of a monument or classical building structure. Near the upper middle section above the “is the entity a expressing an opinion on the C H item?”, there is an illustration of a girl and a boy seated and writing. Near the “Extract hypotheses” box, there is a small illustration depicting a person working with documents or research materials”.Sequential pipeline architecture for SEBI-based KG generation, showing correspondence between components and COPs. Source: Authors' own work