Neural Morphological Analysis: Encoding-Decoding Canonical Segments

Kann, Katharina; Cotterell, Ryan; Schütze, Hinrich

doi:10.18653/v1/d16-1097

Cited by 29 publications

(57 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Wang et al (2016) applied window LSTM model for surface segmentation. Kann et al (2016) improved the results by on canonical segmentation by applying the encoder-decoder RNN framework. Kann et al (2016) achieve the current state-of-the-art for canonical segmentation by re-ranking the output of the encoder-decoder system.…”

Section: Related Workmentioning

confidence: 99%

“…Kann et al (2016) improved the results by on canonical segmentation by applying the encoder-decoder RNN framework. Kann et al (2016) achieve the current state-of-the-art for canonical segmentation by re-ranking the output of the encoder-decoder system. The re-ranking component is a multilayer perception run on the morphemes embeddings (Kann et al, 2016).…”

Section: Related Workmentioning

confidence: 99%

“…The morphemes embedding used for this re-ranking model are calculated using additional information from the Aspell dictionaries. We follow Kann et al (2016) in using the encoder-decoder RNN framework, but we do not use any external resources. Instead of that, we extract and exploit more information from the training corpus.…”

Section: Related Workmentioning

confidence: 99%

“…We follow the notations of Kann et al (2016) in the formalization of the task of canonical segmentation. First, we define two discrete alphabets, Σ of the surface symbols and Σ can of the canonical symbols.…”

Section: Model Descriptionmentioning

confidence: 99%

“…This data is further used to sample 5 splits into 8000 training forms, 1000 development forms and 1000 test forms. Following and Kann et al (2016), we report the results on each of the 5 splits for all three languages. In addition to these sets, we use a manually segmented and glossed corpus of Chintang (Bickel et al, 2004(Bickel et al, -2015, a language that features a high degree of synthesis and free prefix ordering (Bickel et al, 2007).…”

Section: Corporamentioning

confidence: 99%

See 4 more Smart Citations

Neural Sequence-to-sequence Learning of Internal Word Structure

Ruzsics¹,

Samardžić

2017

Proceedings of the 21st Conference on Computational Natural Language Learning (CoNLL 2017)

View full text Add to dashboard Cite

Learning internal word structure has recently been recognized as an important step in various multilingual processing tasks and in theoretical language comparison. In this paper, we present a neural encoder-decoder model for learning canonical morphological segmentation. Our model combines character-level sequence-to-sequence transformation with a language model over canonical segments. We obtain up to 4% improvement over a strong character-level encoderdecoder baseline for three languages. Our model outperforms the previous state-ofthe-art for two languages, while eliminating the need for external resources such as large dictionaries. Finally, by comparing the performance of encoder-decoder and classical statistical machine translation systems trained with and without corpus counts, we show that including corpus counts is beneficial to both approaches.

show abstract

Section: Related Workmentioning

confidence: 99%