使用預訓練編碼器提升跨語言摘要能力;Improving Cross-Lingual Text Summarization using Pretrained Encoder

NCUIR > College of Electrical Engineering & Computer Science > Software Engineer > Electronic Thesis & Dissertation > Item 987654321/83795

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/83795

Title:	使用預訓練編碼器提升跨語言摘要能力;Improving Cross-Lingual Text Summarization using Pretrained Encoder
Authors:	莊家閔;Chuang, Chia-Min
Contributors:	軟體工程研究所
Keywords:	文本摘要;預訓練模型;跨語言處理;Summarization;Pretraining language model;Cross-lingual
Date:	2020-07-31
Issue Date:	2020-09-02 17:07:01 (UTC+8)
Publisher:	國立中央大學
Abstract:	跨語言文本摘要是透過機器將一種語言的文章轉換成另一種語言的摘要，先前的研究大多將該任務以兩步驟方法處理──「先翻譯後摘要」或「先摘要後翻譯」。但是，這兩種方法皆會有翻譯錯誤的問題，且其中的機器翻譯模型難以隨著摘要任務繼續更新微調（fine-tune）。針對上述問題，我們採用預訓練跨語言編碼器以向量表示（represent）不同語言的輸入，將其映射至相同的向量空間。預訓練方法已被廣泛應用在各種自然語言生成任務中，並取得優異的模型表現。此編碼器使得模型在學習摘要能力的過程中，同時保有跨語言能力。本研究中，我們實驗三種不同的微調方法，證明預訓練跨語言編碼器可以學習單詞階層（word-level）的語意特徵。在我們所有的模型組態裡，最優異的模型可在ROUGE-1分數上，超越基準模型3分。 ;Cross-lingual text summarization (CLTS) is the task to generate a summary in one language given a document in a another language. Most of the previous work consider CLTS as two sub-tasks: translate-then-summarize and summarize-then-translate. Both of them are suffered from translation error and the translation system is hard to be fine-tuned with text summarization directly. To deal with the above problems, we utilize a pretrained cross-lingual encoder, which has been demonstrated the effectiveness in natural language generation, to represent text inputs from from different languages. We augment a standard sequence-to-sequence (Seq2Seq) network with our pretrained cross-lingual encoder so as to capture cross-lingual contextualized word representation. We show that the pretrained cross-lingual encoder can be fine-tuned on a text summarization dataset while keeping the cross-lingual ability. We experiment three different fine-tune strategies and show that the pretrained encoder can capture cross-lingual semantic features. The best of the proposed models obtains 42.08 Rouge-1 on ZH2ENSUM datasets [Zhu et al., 2019], significantly improving our baseline model by more than 3 Rouge-1.
Appears in Collections:	[Software Engineer] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	319	View/Open

社群 sharing

Loading...