Open ercanse opened 7 years ago
Update: it seems that no such complicated mechanism is necessary. A list of sentences, separated by newlines, can be passed to the process and it will process them separately, duly indicating which triples correspond to which sentence. The refactoring should make use of this provided functionality.
Currently, for each sentence, a separate Alpino process is created for the extraction of the dependency triples contained in it. Refactor this so that there is only one such process which is used by all annotators in turn, using some mutex mechanism.