The Context (not only) for Humans

Our context considerations will be practically oriented; we will explore the specification of a context scope in the Czech morphological tagging. We mean by morphological tagging/annotation the automatic/manual disambiguation of the output of morphological analysis. The Prague Dependency Treebank (PDT) serves as a source of annotated data. The main aim is to concentrate on the evaluation of the influence of the chosen context on the tagging accuracy. 1. English and Czech Tagging Experiments The corpus-based approaches determine the amount of human work involved in the NLP tasks on the building of training data and on the coming up with an algorithm giving results as precise as possible. The ideas of context specification cannot be left out in the formulation of the algorithm. The scope of context must be specified according to the character of the particular NLP task. As we consider the nature of context from the perspective of the tagging application, an elementary unit we process is a word token. In general, there is no strict rule saying how many preceding and following word tokens we should look at to be sure that we tag the word token properly. Thus, let us have a look at the empirical experience. STRATEGY TAGGER TRAINING TA