知能コミュニケーション研究室のJohanes Effendiさん(前期課程2年)らが、国際会議Oriental COCOSDA 2017においてOriental COCOSDA 2017 ITN Best Student Paper Awardを受賞しました。
Johanes Effendi et al. were awarded the Oriental COCOSDA 2017 ITN Best Student Paper Award at the 20th Oriental COCOSDA conference.
(2017/11/3)
|
Oriental COCOSDA（COCOSDAオリエンタルチャプター）は、アジア言語の音声言語コーパスの作成や利用、普及、さらに音声認識/合成システムの評価方法について、意見交換や情報共有を行う場を提供し、その地域特性について議論することを目的とした、アジア言語の音声研究を推進する国際会議です。 Oriental COCOSDA 年次国際会議は今年で20回目を迎え、2017年11月1日から3日までソウルにて開催されました。 |
 |
- 受賞者/著者 Awardee/Authors:
Johanes Effendi, Sakriani Sakti, Satoshi Nakamura
- 受賞テーマ Research theme:
"CREATION OF A MULTI-PARAPHRASE CORPUS BASED ON VARIOUS ELEMENTARY OPERATIONS"
Paraphrases resemble monolingual translations from a sourcesentence into other sentences that must preserve the originalmeaning. To build automatic paraphrasing, a collection ofparaphrased expressions is required. However, manually collectingparaphrases is expensive and time-consuming. Mostexisting paraphrases corpora cover only one-to-one parallelsentences and neglect the fact that possible variants of paraphrasescan be generated from a single source sentence. Themanipulation applied to the original sentences is also difficultto track. Furthermore, a single corpus is mostly dedicated toa single application that is not reusable in other applications.In this research, we construct a paraphrase corpus based onvarious elementary operations (reordering, substitution, deletion,insertion) in a crowdsourcing platform to generate multi-paraphrasesentences from a source sentence. These elementaryparaphrase operations can be utilized for various applications(i.e., deletion for summarization and reordering formachine translation). Our evaluations show the richness andeffectiveness of our created corpus.
- 受賞者のコメント Awardee's voice
This corpus proposed non-typical way to do paraphrasing by creating multiple paraphrases that follows four elementary operations. This creation method delivers a corpus that is decomposable based on what suitable operations are needed in a particular task. We hope that this corpus will be useful for further improving language technologies in the future. Furthermore, I would like to express my sincere gratitude to Prof. Satoshi Nakamura, Assistant Prof. Sakriani Sakti, and all members of Augmented Human Communication Laboratory for their valuable comment, feedback, and support during conducting this research. May this award is encouraging us to improve our research even more in the future.
- 外部リンク
Oriental COCOSDA 2017のページ: http://ococosda2017.org/