Implementing the multilingual Semantic Web vision requires the transformation of unstructured data in multiple languages from the Document Web into structured data for the multilingual Web of Data. We present the multilingual version of FOX[2], a knowledge extraction framework which supports this migration by providing named entity recognition based on ensemble learning for five languages (German, English, Spanish, French and Dutch). The framework is open source, freely available and ready to use by the community via a RESTful web service.
Research papers:
Ensemble Learning for Named Entity Recognition [1]
- Ensemble Learning of Named Entity Recognition Algorithms using Multilayer Perceptron for the Multilingual Web of Data” [under review]
- Ensemble Learning of Named Entity Recognition Algorithms using Multilayer Perceptron for the Multilingual Web of Data” [1a]
Thus, we push forward this version of the multilingual Web of Data with a multilingual state of the art system. FOX provides the results in the NIF[3] and enriches them with provenance information by using the PROV-O ontology[4]. It also links results with the integrated NED tool Agdistis[5] to the DBpedia knowledge base. We extended the framework with the new versions of Agdistis to support a better entity linking. In the near future, we plan to integrate more NER tools in the framework’s pipeline with the aim of improving the performance, particularly for languages that only have a few tools integrated in their current version, like Dutch, and for languages that are currently missing in the pipeline, e.g. Italian.
[1] <link https: svn.aksw.org papers iswc_el4ner public.pdf>svn.aksw.org/papers/2014/ISWC_EL4NER/public.pdf
[1a] <link https: svn.aksw.org papers kcap_fox public.pd>svn.aksw.org/papers/2017/KCAP_FOX/public.pd
[2] <link http: fox.aksw.org>fox.aksw.org
[3] <link http: persistence.uni-leipzig.org nlp2rdf>persistence.uni-leipzig.org/nlp2rdf
[4] <link https: www.w3.org tr prov-o>www.w3.org/TR/prov-o
[5] <link https: github.com dice-group agdistis>github.com/dice-group/AGDISTIS
