Adversarial Alignment of Multilingual Models for Extracting Temporal Expressions from Text

Although temporal tagging is still dominated by rule-based systems, there have been recent attempts at neural temporal taggers. However, all of them focus on monolingual settings. In this paper, we explore multilingual methods for the extraction of temporal expressions from text and investigate adversarial training for aligning embedding spaces to one common space. With this, we create a single multilingual model that can also be transferred to unseen languages and set the new state of the art in those cross-lingual transfer experiments.

PDF Abstract WS 2020 PDF WS 2020 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Temporal Tagging Basque TimeBank Lange et al. F1 47.87 # 1
Temporal Tagging Catalan TimeBank 1.0 Lange et al. F1 64.21 # 1
Temporal Tagging French Timebank Lange et al. F1 62.58 # 1
Temporal Tagging KRAUTS Lange et al. F1 66.53 # 1
Temporal Tagging Spanish TimeBank 1.0 Lange et al. F1 79.55 # 1
Temporal Tagging TempEval-3 Lange et al. F1 74.8 # 1
Temporal Tagging TimeBankPT Lange et al. F1 75.47 # 1