Tuesday, November 3, 2009

Semantics For Document Management (NTI) - Fourth demo implemented

Yerbabuena has published the fourth NTI version (Nuxeo Thinks Itself) of the Semantics For Nuxeo Proyect 2010.

In this fourth sprint have been implemented these new functionalities:

  • Possibility to add annotations about localizations through Google Maps API.

  • DBPedia and OpenCalais integration as additional semantic annotations engines over the documents.

  • Progressive learning in identifying the types of documents.

  • Automatic start of actions depending on the type of documents.

  • Automatic and preiodic background reasoning work.

First of all, Nuxeo has been integrated with Google Maps API so that locations can be added as annotations directly from the map view.

Furthermore, we have expanded the semantic annotation engine, allowing automatic annotation of documents. Such annotations are made through OpenCalais and DBpedia. Once you create a document, its content is sent by a thread, in background, to the OpenCalais and DBPedia services, returning a series of tags that are added as annotations to the document. Nuxeo is also integrated with the Google API translator, as both DBpedia and OpenCalais offer better services if the text is written in English. Once the thread has finished, it displays a message informing the user and throws an event that is picked up by Nuxeo and added to the history tab.

Learning the types of documents allows Nuxeo can automatically identify and classify all documents, for which it must have been trained previously. This way you can "teach" the semantic service to be able to identify any type of document.

Furthermore, we have developed an extension point that allows automatic startup of actions depending on the type of document. These actions could be a workflow or any other type of action runnable in Nuxeo. Defining and setting up the parameters for this action is done via an XML which is then parsed at deployment time.

Finally, we have implemented an additional thread within the semantic service that handles the task of reasoning automatically, periodically and in background, transparently to the user.


No comments:

Post a Comment