Extension: EntityFCS

Entity Search and Integration

Resource and Materials

• Demo:
  • @text+: https://tppssi-demo.saw-leipzig.de/
    • Enable Entity Search in settings
    • Switch to Advanced Search (Only supported there)
      • Trigger entity suggestions by typing @ in the query input field (uses lobid-API for GND entities IDs)
    • Switch to Results View, enable annotation layer view
      • Tokens or Spans may be underlined where Entities can be annotated, by hovering an info box will pop up that uses the GND to fetch meta information about an entity by using its ID

Motivation

Searching using an entity ID

• cover alternative forms (spelling mistakes, historical forms)
• synonyms, synsets
• co-references/mentions: "he"/"she"/"they", "the person", ...
• disambiguation (homonyms; same spelling but different person or different word usage based on context)

Questions

• Which IDs to support
• Entity ID resolution (due to different vocabularies)
  • on endpoint or client (FCS Aggregator, search support) side
• Idea: FCS client (FCS Aggregator) with keyword expansion using entity ID to generate list of word forms for native search

Implementation and Extension of FCS

• Advanced Search with new Layer Type Identifier of entity that can be used to annotate tokens/spans with entity IDs (prefixed or full URIs)
  • FCS-QL can then search with the entity field type, custom identifier may be used to distinguish between different entity layers
  • Q: Should entity IDs be fully resolved URIs or can then be short identifiers?
    Q: If short identifiers (e.g., a number or string) should they be prefixed with the knowledge base identifier (e.g., wikidata:, gnd:, geonames:) or are they distinguishable using the layer with its ID? Searching in the generic entity layer would make using prefixed entity IDs better for disambiguation. Q: Are prefixed variants easier for users for searching? (also for input support?)
• in Vuetify FCS Frontend (demo) with custom search input support (ID suggestions) and visualization (see section Resource and Materials)
• Entity search "mode" in FCS Aggregator to generate search queries (using IDs) for both Advanced Search and Lexical Search (see LexFCS - FCS for Lexical Resources) for aggregated results over multiple search query types