Wikimeta is a semantic tagging and annotation architecture that allows you to incorporate semantic knowledge within your documents, website, content management system. Wikimeta is fully compatible with the LinkedData Network, including DBPedia, Geonames, CIAFactbook, Wikicompanies and more. Wikipedia is fully multilingual, and can annotate text content in English (beta), and French (beta).
Currently, Wikimeta can produce Part Of Speech Annotation, Named entities annotation (LOC, PERS, ORG, TIME, DATE, PROD) and associate those with semantic tags. It's high degre of precision allows It to be used for exemple in the BIO-NLP co-reference resolution task to supply a dedicated labeling tool.
Wikimeta is strictly multilingual (each language is supported by a specific model) and is currently available in French and English.
Wikimeta API is strictly compliant with REST, XML and JSON formats. The API is free of use with a bandwith limit (expendable for academic needs).
The data set used to train the NLP tools of Wikimeta are derived from wikipedia and are also available for download to build your own applications, gazeteers or training corpora (see www.nlgbase.org).
We plan to introduce the co-reference resolution developed for ConNLL shared task 2011 in Wikimeta at the beginning of 2012.
We currently work on concept labeling including those involved in BIO-NLP.