I have configured an enhanement chain in stanbol including tika,language Detection, Custom Named Entity Recogniser and OpenNLP Sentence Detector.
I am clearly getting the NER and lang-detect output but the setence detection engine dosent seem to work.
Am I missing something here?
Instead of using opennlp-sentence you should use opennlp-token. The sentence detection is useless because the model employed for the training model is only tokenized (and of course labelled with the entities). In output you won't ever see how the input is tokenized (or segmented into sentences), but it will show you the context from where it infers the output data.
