Releases: nheist/CaLiGraph
Bugfix-Release improving NT-file formatting
Bugfix release for ontology consistency
With this release we deploy several bugfixes for issues that made the ontology inconsistent. In particular, class disjointnesses are treated correctly again.
Extracting *disambiguated* entities from all listings in Wikipedia
With the release 3.0, we implement the changes described in our ESWC'23 paper "NIL-Aware Scalable Transformer-based Entity Linker" (https://arxiv.org/pdf/2303.04426.pdf). We deploy an approach that identifies and disambiguates novel entities and facts from tables and enumerations all over Wikipedia. For detailed statistics of the new version, visit http://caligraph.org/.
Extracting entities from all listings in Wikipedia
With the release 2.0 we implement the changes described in our WWW'21 paper "Information Extraction from Co-Occurring Similar Entities" (https://arxiv.org/pdf/2102.05444). In particular, we extract novel entities and facts from tables and enumerations all over Wikipedia. For detailed statistics of the new version, visit http://caligraph.org.
Improved Subject Entity Recognition with spaCy 3.0 and BERT
We improved the recognition of subject entities with a BERT-based tagger and improved language models from spaCy v3.0.
Improved formatting in serialised files
Fixed an issue that may lead to incorrectly formatted n-triple files.
CaLiGraph for DBpedia2020
Updated version of CaLiGraph that is based on recent files of DBpedia. Additionally, many other fixes for correctness and stability.
CaLiGraph Version 1.0 Release
Initial release of the CaLiGraph extraction code.