论文信息 - Deep Learning for Automatic Diacritics Restoration in Romanian

Deep Learning for Automatic Diacritics Restoration in Romanian

In this paper we address the issue of automatic diacritics restoration (ADR) for Romanian using deep learning strategies.We compare 6 separate architectures with various mixtures of recurrent and convolutional layers. The input consists in sequences of consecutive words stripped of their diacritic symbols. The network’s task is to learn to restore the diacritics for the entire sequence. No additional linguistic or semantic information is used as input to the networks.The best results were obtained with a CNN-based architecture and achieved an accuracy of 97% at word level. At diacritic-level the accuracy of the same architecture is 89%.

[1] Jürgen Schmidhuber,et al. Long Short-Term Memory , 1997, Neural Computation.

[2] Jan Hajic,et al. Diacritics Restoration Using Neural Networks , 2018, LREC.

[3] Quoc V. Le,et al. Sequence to Sequence Learning with Neural Networks , 2014, NIPS.

[4] Michel Simard. Automatic Insertion of Accents in French Text , 1998, EMNLP.

[5] Horia Cucu,et al. A robust diacritics restoration system using unreliable raw text data , 2014, SLTU.

[6] Verginica Barbu Mititelu,et al. CoRoLa ― The Reference Corpus of Contemporary Romanian Language , 2014, LREC.

[7] Cheol-Young Ock,et al. Diacritics Restoration in Vietnamese: Letter Based vs. Syllable Based Model , 2010, PRICAI.

[8] Rada Mihalcea,et al. Letter Level Learning for Language Independent Diacritics Restoration , 2002, CoNLL.

[9] C. Negrescu,et al. AUTOMATIC DIACRITIC RESTORATION FOR A TTS-BASED E-MAIL READER APPLICATION , 2008 .

[10] Rada Mihalcea,et al. Diacritics Restoration: Learning from Letters versus Learning from Words , 2002, CICLing.