Bridging the domain gap in cross-lingual document classification

16 Sep 2019  ·  Guokun Lai, Barlas Oguz, Yiming Yang, Veselin Stoyanov ·

The scarcity of labeled training data often prohibits the internationalization of NLP models to multiple languages. Recent developments in cross-lingual understanding (XLU) has made progress in this area, trying to bridge the language barrier using language universal representations... However, even if the language problem was resolved, models trained in one language would not transfer to another language perfectly due to the natural domain drift across languages and cultures. We consider the setting of semi-supervised cross-lingual understanding, where labeled data is available in a source language (English), but only unlabeled data is available in the target language. We combine state-of-the-art cross-lingual methods with recently proposed methods for weakly supervised learning such as unsupervised pre-training and unsupervised data augmentation to simultaneously close both the language gap and the domain gap in XLU. We show that addressing the domain gap is crucial. We improve over strong baselines and achieve a new state-of-the-art for cross-lingual document classification. read more

PDF Abstract

Datasets


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Cross-Domain Document Classification Amazon cn (Dianping train) XLMft UDA Error rate 7.74 # 1
Cross-Domain Document Classification Amazon en (Yelp train) XLMft UDA Error rate 7.57 # 1
Cross-Domain Document Classification Dianping (Amazon cn train) XLMft UDA Error rate 4.64 # 1
Cross-Lingual Sentiment Classification Dianping (Yelp train) XLMft UDA Error rate 4.64 # 1
Cross-Lingual Sentiment Classification MLDoc Zero-Shot English-to-Chinese XLMft UDA Error rate 7.74 # 1
Cross-Lingual Document Classification MLDoc Zero-Shot English-to-Chinese XLMft UDA Accuracy 93.32 # 1
Cross-Lingual Sentiment Classification MLDoc Zero-Shot English-to-French XLMft UDA Error rate 5.95 # 1
Cross-Lingual Document Classification MLDoc Zero-Shot English-to-French XLMft UDA Accuracy 96.05 # 1
Cross-Lingual Document Classification MLDoc Zero-Shot English-to-German XLMft UDA Accuracy 96.95% # 1
Cross-Lingual Sentiment Classification MLDoc Zero-Shot English-to-German XLMft UDA Error rate 6.12 # 1
Cross-Lingual Document Classification MLDoc Zero-Shot English-to-Russian XLMft UDA Accuracy 89.7 # 1
Cross-Lingual Document Classification MLDoc Zero-Shot English-to-Spanish XLMft UDA Accuracy 96.8 # 1
Cross-Domain Document Classification Yelp (Amazon en train) XLMft UDA Error rate 3.34 # 1

Methods


No methods listed for this paper. Add relevant methods here