On the Frailty of Universal POS Tags for Neural UD Parsers

We present an analysis on the effect UPOS accuracy has on parsing performance. Results suggest that leveraging UPOS tags as features for neural parsers requires a prohibitively high tagging accuracy and that the use of gold tags offers a non-linear increase in performance, suggesting some sort of exceptionality. We also investigate what aspects of predicted UPOS tags impact parsing accuracy the most, highlighting some potentially meaningful linguistic facets of the problem.

[1]  Timothy Dozat,et al.  Deep Biaffine Attention for Neural Dependency Parsing , 2016, ICLR.

[2]  Karin M. Verspoor,et al.  An Improved Neural Network Model for Joint POS Tagging and Dependency Parsing , 2018, CoNLL.

[3]  Noah A. Smith,et al.  Improved Transition-based Parsing by Modeling Characters instead of Words with LSTMs , 2015, EMNLP.

[4]  Yu Zhang,et al.  Is POS Tagging Necessary or Even Helpful for Neural Dependency Parsing? , 2020, NLPCC.

[5]  Benjamin Heinzerling,et al.  BPEmb: Tokenization-free Pre-trained Subword Embeddings in 275 Languages , 2017, LREC.

[6]  Noah A. Smith,et al.  Many Languages, One Parser , 2016, TACL.

[7]  Joakim Nivre,et al.  From Raw Text to Universal Dependencies - Look, No Tags! , 2017, CoNLL.

[8]  Mary Dalrymple.  How much can part-of-speech tagging help parsing? , 2006, Nat. Lang. Eng..

[9]  Cheikh M. Bamba Dione,et al.  Developing Universal Dependencies for Wolof , 2019, Proceedings of the Third Workshop on Universal Dependencies (UDW, SyntaxFest 2019).

[10]  Yue Zhang,et al.  NCRF++: An Open-source Neural Sequence Labeling Toolkit , 2018, ACL.

[11]  Joakim Nivre,et al.  82 Treebanks, 34 Models: Universal Dependency Parsing with Multi-Treebank Models , 2018, CoNLL.

[12]  Meishan Zhang,et al.  Joint POS Tagging and Dependence Parsing With Transition-Based Neural Networks , 2017, IEEE/ACM Transactions on Audio, Speech, and Language Processing.

[13]  Haitao Liu,et al.  Dependency Distance as a Metric of Language Comprehension Difficulty , 2008 .

[14]  Timothy Dozat,et al.  Stanford’s Graph-based Neural Dependency Parser at the CoNLL 2017 Shared Task , 2017, CoNLL.

[15]  Atro Voutilainen.  Does tagging help parsing? A case study on finite state parsing , 1998 .

[16]  Yuan Zhang,et al.  Stack-propagation: Improved Representation Learning for Syntax , 2016, ACL.

[17]  Vikas Raunak.  Simple and Effective Dimensionality Reduction for Word Embeddings , 2017 .

[18]  Joakim Nivre,et al.  An Investigation of the Interactions Between Pre-Trained Word Embeddings, Character Models and POS Tags in Dependency Parsing , 2018, EMNLP.

[19]  Hai Zhao,et al.  Joint Learning of POS and Dependencies for Multilingual Universal Dependency Parsing , 2018, CoNLL.

[20]  Carlos G'omez-Rodr'iguez,et al.  Distilling Neural Networks for Greener and Faster Dependency Parsing , 2020, IWPT.

[21]  Daniel Zeman,et al.  CoNLL 2017 Shared Task - Automatically Annotated Raw Texts and Word Embeddings , 2017 .

[22]  Prakhar Gupta,et al.  Learning Word Vectors for 157 Languages , 2018, LREC.

[23]  Ramadan Alfared.  POS Taggers and Dependency Parsing , 2012 .

[24]  David Vilares,et al.  Viable Dependency Parsing as Sequence Labeling , 2019, NAACL.