Leveraging Large Amounts of Weakly Supervised Data for Multi-Language Sentiment Classification
It addresses the limited training data problem for non-English languages in sentiment analysis, offering a novel approach that is incremental in method but impactful for domain-specific applications.
The paper tackles multi-lingual sentiment classification in short texts by leveraging weakly supervised data without requiring English correspondence, achieving state-of-the-art performance on benchmarks like SemEval-2016.
This paper presents a novel approach for multi-lingual sentiment classification in short texts. This is a challenging task as the amount of training data in languages other than English is very limited. Previously proposed multi-lingual approaches typically require to establish a correspondence to English for which powerful classifiers are already available. In contrast, our method does not require such supervision. We leverage large amounts of weakly-supervised data in various languages to train a multi-layer convolutional network and demonstrate the importance of using pre-training of such networks. We thoroughly evaluate our approach on various multi-lingual datasets, including the recent SemEval-2016 sentiment prediction benchmark (Task 4), where we achieved state-of-the-art performance. We also compare the performance of our model trained individually for each language to a variant trained for all languages at once. We show that the latter model reaches slightly worse - but still acceptable - performance when compared to the single language model, while benefiting from better generalization properties across languages.