Paper Title
A Review of Financial Sentiment Analysis

Abstract
The increased number of users’ financial posts on the Internet provides an excellent opportunity for individuals and companies to analyze the stock market and predict its variations to make wise investment decisions. Financial sentiment analysis (FSA) is defined as the identification and extraction of sentiment orientation in financial content. This paper reviews the literature to explore the potential of existing financial sentiment analysis techniques. It covers primary financial dataset sources, a lexicon-based method, a machine learning approach, and a deep learning approach as applied in FSA tasks. Moreover, the evaluation metrics used to evaluate the financial sentiment models' performance, including accuracy, F1 score, Mean Square Error, R2, and cosine similarity, are discussed. This paper also highlights the potential challenges and future directions in the financial sentiment analysis domain, such as the problem of domain adaptation, the quality of sentiment labeling, contextualizing word embedding, multi-lingual sentiment analysis, and the attention mechanism for aspect extraction. Keywords - Sentiment Analysis, Financial Sentiment Analysis, Lexicon-Based, Deep Learning, Machine Learning, Stock Market Prediction