Temporal Tagging Tasks

•extraction of temporal expressions •normalization of temporal expressions Most approaches • focus on English • focus on news-style texts Main challenges [1] •different domains, different challenges •normalizing non-explicit expressions, e.g., today, Monday, next week, July •only few languages addressed so far Multilingual temporal tagging •annotated corpora in several languages • few multilingual temporal taggers •earlier works on automatic extensions to new languages less successful So far: temporal tagging of many languages never addressed! HeidelTime Key features [1] • rule-based system •multilingual and domain-sensitive Architecture • language-independent, domain-specific normalization strategies • language-dependent resources • required: sentence, token, pos information ResourcesSource Code HeidelTime’s Language Resources (i) pattern files (ii) normalization files / / reMonthLong [Ee]ner