Fin-STance: A Novel Deep Learning-Based Multi-Task Model for Detecting Financial Stance and Sentiment

Stance detection has been gaining popularity in text mining and information retrieval-based research. Recognition of sentiment also plays an important role in the analysis of text whether the underline stances are in favor or against for a particular domain. However, there is extensive research in this direction, but considering its impact on the financial text has not been explored yet. This paper presents a new deep learning based multi-task approach for financial stance detection and sentiment detection. A new model called Fin-STance is introduced which performs two-classification tasks as stance detection and sentiment detection based on the financial data. It constitutes of an input, embedding, followed by shared and task-specific layers. The empirical evaluation of this study is conducted on a newly created dataset and it shows impressive results. Also, this study receives better results with relevant compared methods.