Hello, the English NER models were trained on MUC6 and MUC7 data, plus some manually labeled data.
No, it is not possible to add any training data to an existing model. HTH, Jörn On Wed, 2014-04-16 at 19:20 -0400, Marco Farrugia wrote: > I realize the training dataset is under copyright, and not easily > redistributable. But I'm still interested in knowing which data set was > used. > > A second question, is there support for continuing to train a model with > additional training data? As in loading the available model binary, and > supplementing with additional training. (Apologies for needing to ask, as I > am still somewhat unfamiliar with the OpenNLP source).
