- From: Michele Catasta <michele.catasta@epfl.ch>
- Date: Wed, 23 Oct 2013 01:48:12 +0100
- To: public-lod@w3.org
- Cc: Alberto Tonon <alberto.tonon@exascale.info>, Gianluca Demartini <demartini@exascale.info>, Philippe Cudre-Mauroux <phil@exascale.info>
TRank is a pipeline that, given a textual/HTML document as input, performs named-entity recognition, entity linking/disambiguation, and entity type ranking/selection from a variety of type hierarchies including DBpedia, YAGO, and schema.org. TRank has been nominated as best paper at ISWC2013. We have now released TRank open-source for others to use: https://github.com/MEM0R1ES/TRank It provides good test coverage, continuous build, and exhaustive documentation. You can use it as is, or easily integrate your own entity type ranking algorithm to compare against or to build on top of TRank. Bug reports and pull requests are welcome! We also recommend to watch/star the GitHub repository, as we will be releasing soon the MapReduce implementation of TRank. -- Best, Michele
Received on Wednesday, 23 October 2013 09:29:03 UTC