Experiments in tagger combination: arbitrating, guessing, correcting, suggesting
Giuseppe M. Attardi, A. Fuschetto, F. Tamberi, Maria Simi, Eva Maria Vecchi · CINECA IRIS Institutial research information system (University of Pisa) · 2009
The paper reports experiments with several strategies of tagger combination, using two well known taggers, TreeTagger and Hunpos. The most successful experiment,which achieved the best score in the Evalita 2009 Open Task, is a hybrid solution combining an easiest first iterative strategy with hand-written arbitration rules and a quality lexicon. The system used in the Closed Task uses a similar combination of two variants of Hunpos, together with correction/guessing rules.