Hi, I am a computer science student from Germany. While I do not have so much coding experience, I really NEVER got in touch with machine learning so far, but need it for my current project. So by advice of a fellow student I came across your maxent implementation and the nice Howto, which helped me a lot to understand and use your implementation, thank you very much for that!
Now to get to my problem: My event stream sometimes contains events, which consist of an outcome and a context String array containing only ONE String! When this event gets processed, a null pointer exception occurs and I managed to trace back, why this happens: The method opennlp.model.AbstractDataIndexer.sortAndMerge(List,boolean) returns, if numEvents <= 1 resulting in 'contexts' to stay null. Now my question: Is there an important reason, why numEvents has to be <=1 ? Which dark things happen, if you change it to numEvents<1 ? In a first test, changing it effected in not crashing and also the prediction value looked fine, but it still feels like bad hacking and not knowing what actually happens internally. I thank you very much in advance for any advice in this case! Best regards, Daniel -- NEU: FreePhone - kostenlos mobil telefonieren und surfen! Jetzt informieren: http://www.gmx.net/de/go/freephone
