FOX - Federated knOwledge eXtraction Framework

FOX is a framework for RDF extraction from text based on Ensemble Learning. It makes use of the diversity of NLP algorithms to extract entities with a high precision and a high recall. Moreover, it provides functionality for Keyword Extraction and Relation Extraction. 

FOX relies on ensemble learning and combines the results of several tools by using decision-tree-based algorithms. In addition to detecting entities, FOX uses AGDISTIS to disambiguate these entities against DBpedia. In contrast to lookup-based apprioaches, our framework can also detect resources that are not in DBpedia. These are assigned their own URIs. FOX was evaluated with the datasets available here. The output of the evaluation can be found here.

Fox can be used in many ways:

  1. You can choose to use the running FOX service, which we try to alive as much as we can. An API to address FOX programmatically can be found the following link.

  2. You can choose to simply run your own FOX. FOX is open-source (GPL v3 license). Please feel free to fork FOX on Github.

Project Team

527efb333