Efficient Higher-Order CRFs for Morphological Tagging

By Thomas Müller, Helmut Schmid, Hinrich Schütze. 2013

222 citations recorded by the source corpus.

Training higher-order conditional random fields is prohibitive for huge tag sets. We present an approximated conditional random field using coarse-to-fine decoding and early updating. We show that our implementation yields fast and accurate morphological taggers across six languages with different morphological properties and that across languages higher-order models give significant improvements over 1-order models.

Subfields: Artificial Intelligence, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

Citations in the corpus

Recurrent Neural Network Grammars (2016, background): “…similarly as in (Dyer et al., 2015); they are predicted for English with the Stanford Tagger (Toutanova et al., 2003) and Chinese with Marmot (Mueller et al., 2013). inatively trained variant of our parser (§4.4) fulfills these requirements: sequences of actions can be sampled using a…”
What do Neural Machine Translation Models Learn about Morphology? (2017, background): “Please refer to Mueller et al. (2013) for representative POS and morphological tagging accuracies.”
Finding Function in Form: Compositional Character Models for Open Vocabulary Word Representation (2015, background): “As morphology is a strong indicator for syntax in many languages, a much effort has been spent engineering features (Nakagawa et al., 2001; Mueller et al., 2013).”
Improved Transition-based Parsing by Modeling Characters instead of Words with LSTMs (2015, methodology): “For Turkish, we show the results of Nivre et al. (2006a) which also carried out a careful manual morphological feature selection.”
Phrase-Based SMT for Finnish with More Data, Better Models and Alternative Alignment and Translation Tools (2016, methodology): “The readings given by OMorFi are combined with predictions of the MarMoT CRF-based tagger (Mueller et al., 2013), and the data is subsequently parsed using the mate-tools data-driven dependency parser (Bohnet, 2010).”
Predicting Prosodic Prominence from Text with Pre-trained Contextualized Word Representations (2019, methodology): “…(BiLSTM) (Hochreiter and Schmidhuber, 1997) • Minitagger (SVM) (Stratos and Collins, 2015) + GloVe (Pennington et al., 2014) • MarMoT (CRF) (Mueller et al., 2013) • Majority class per word The models were selected so that they cover a wide variety of different architectures from…”
From POS tagging to dependency parsing for biomedical event extraction (2018, methodology): “We consider the following: • MarMoT [14] is a well-known generic CRF framework as well as a leading POS and morphological tagger.”
Mimicking Word Embeddings using Subword RNNs (2017, background): “M¨uller et al. (2013) do the same for six other languages.”
Coreference and Coherence in Neural Machine Translation: A Study Using Oracle Experiments (2018, methodology): “Therefore, we used MarMoT (Mueller et al., 2013) to ﬁlter out these instances.”