Sentiment Analysis Using Aligned Word Embeddings for Uralic Languages
This enables state-of-the-art sentiment analysis for endangered languages, requiring only a dictionary to a majority language, though it is incremental as it adapts existing methods.
The paper tackled sentiment analysis for four endangered Uralic languages by translating and aligning word embeddings from English, then applying a neural network trained on English data. The method achieved at least 56% accuracy on newly annotated corpora for each language.
In this paper, we present an approach for translating word embeddings from a majority language into 4 minority languages: Erzya, Moksha, Udmurt and Komi-Zyrian. Furthermore, we align these word embeddings and present a novel neural network model that is trained on English data to conduct sentiment analysis and then applied on endangered language data through the aligned word embeddings. To test our model, we annotated a small sentiment analysis corpus for the 4 endangered languages and Finnish. Our method reached at least 56\% accuracy for each endangered language. The models and the sentiment corpus will be released together with this paper. Our research shows that state-of-the-art neural models can be used with endangered languages with the only requirement being a dictionary between the endangered language and a majority language.