[CL-AFF shared task] multi-label text classification using an emotion embedding model

In this paper, we propose a deep learning model-based ap-proach that combines a language embedding model and an emotion em-bedding model in the classification of text for the CL-AFF Shared Task 2020. The task aims to predict the disclosure and supportiveness labels of the comments (to the posts) in the OffMyChest dataset which consists of a small labeled dataset and a large unlabeled dataset. We investigate the effectiveness of the BERT, Glove, and Emotional Glove embedding models, to represent the text for label prediction. We also propose to use the original posts in the dataset as contextual information. We evaluated our approach and report the results.