AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |
Back to Blog
Elsawin final code generator5/3/2023 ![]() ![]() ![]() You can see the WWW 2019 (known as The Web Conference) paper “ Emoji-Powered Representation Learning for Cross-Lingual Sentiment Classification” for more details. It translates the labeled English data into the target language through an off-the-shelf machine translation system, represent the pseudo parallel texts with the pre-trained language-specific models, and use an attention model to train the final sentiment classifier. This step is also conducted separately for both the source and the target languages. In a distant-supervised way, it uses emojis as complementary sentiment labels to transform the word embeddings into a higher-level sentence representation that encodes rich sentiment information via an emoji-prediction task through an attention-based stacked bi-directional LSTM model. It uses large-scale Tweets to learn word embeddings through Word2Vec of both the source and the target languages in an unsupervised way. ![]() The workflow of ELSA consists of the following phases: ELSA is an emoji-powered representation learning framework for cross-lingual sentiment classification. ![]()
0 Comments
Read More
Leave a Reply. |