Anders Søgaard

Papers

1. Ellipsis Resolution as Question Answering: An Evaluation (2021)
   The paper treats ellipsis resolution as a reading-comprehension problem. The elided clause becomes a question, and the antecedent becomes the answer span in the preceding text. The authors train BERT-based models on sluice ellipsis and VP ellipsis, first alone and then jointly with question-answering and coreference…
2. Historical Text Normalization with Delayed Rewards (2019)
   Sequence-to-sequence normalisers are usually trained with token-level maximum likelihood. The paper fine-tunes such a model with policy gradient, so that the training signal is the exact match of the whole normalisation, a delayed reward. Six historical datasets are used. Fine-tuning reduces the error by 20% for…
3. Few-Shot and Zero-Shot Learning for Historical Text Normalization (2019)
   The paper evaluates 63 multi-task configurations of an encoder-decoder normaliser on ten datasets in eight languages.
4. Multi-task learning for historical text normalization: Size matters (2018)
   The paper asks when multi-task learning helps historical normalisation. It tests an encoder-decoder model with and without auxiliary tasks on 10 datasets from 8 languages, from German and English to Hungarian, Icelandic, Slovene, Spanish, Portuguese and Swedish. Most datasets either always gain or always lose from…
5. Learning attention for historical text normalization by learning to pronounce (2017)
   The paper treats historical normalisation as a character-level transduction with an encoder-decoder RNN, so no explicit character alignment is needed.
6. Learning a POS tagger for AAVE-like language (2016)
   Newswire POS taggers do poorly on subtitles, hip-hop lyrics and tweets in African-American Vernacular English.
7. Improving historical spelling normalization with bi-directional LSTMs and multi-task learning (2016)
   The paper frames historical spelling normalisation as character-level sequence labelling and uses a bi-directional LSTM for it.
8. An Empirical Study of Differences between Conversion Schemes and Annotation Guidelines (2013)
   Many dependency treebanks are converted from constituency treebanks, and different conversion schemes make different choices about heads and relation labels.

Co-authors (9)

• Marcel Bollmann ×5
• Joachim Bingel ×2
• Anna Jørgensen
• Daniel Hardt
• Dirk Hovy
• Matthew Lamm
• Natalia Korchagina
• Rahul Aralikatte
• Simon Flachs

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/authors/anders-sogaard/