Lexicon Filter Sample Clauses

Lexicon Filter. Before human post-editors select the entries which they really want to keep, a possibility has been created to remove unwanted term candidates. Such entries could be:  Candidates which are already known; they need not be reviewed a second time.  Candidates which do not belong to a specific domain (e.g. automotive); the filter then would be a general-domain lexicon, letting pass only narrow-domain words.  Candidates which contain certain stopwords (like en „large’).  Candidates which are known to be irrelevant. The system offers the option to apply a filter which blocks this kind of entries. Users would provide the filter data themselves; only non-matching entries pass the lexicon filter. The basic P2G tool is described in (Thurmair/Aleksić 2012). 10 These heuristics for truecasing and for lemma creation leave room for errors, e.g. in cases where the prenominal adjective is in comparative form (de „der frühere Präsident‟ -> *„der frühe Präsident‟), or in cases where the head should be in plural (en „facts & figures‟ -> *„fact & figure‟). However, they show the best performance overall.