A pipeline for structuring Nordisk familjebok editions extracts headwords at 97.8% F1, classifies at 93.4% F1, matches across editions at 93% precision, and links to Wikidata at 85% precision with 16.5% recall.
It consists of four major steps, notably an automated headword extraction, where we achieved an F1 scoreof97.8%andanentitytypeclassificationwith an F1 score of 93.4%
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
ATLAS: Article Tracking, Linking, and Analysis of Swedish Encyclopedias
A pipeline for structuring Nordisk familjebok editions extracts headwords at 97.8% F1, classifies at 93.4% F1, matches across editions at 93% precision, and links to Wikidata at 85% precision with 16.5% recall.