Stock Price Prediction using Social News Sentiment Analysis

The application of Artificial Intelligence (AI) in the financial sectorhas been gaining interest since the 1990s, especially in the last period where there have been clear technological improvements that have en-abled the use of increasingly sophisticated techniques. One of the most widely discussed and studied issues is the forecastingof market trends, so that it can support investment and stock trading. Where techniques from statistics were previously used, solutions thatmake use of Machine Learning (ML) and Deep Learning (DL) modelsare now increasingly being adopted, achieving better performance. However, in recent years there has been an increased focus on refiningthese methodologies, trying to add additional information other than those coming from the market. Specifically in the literature, attemptsare being made to find correlations between market trends and people’s sentiment, which can be derived from the evaluation of news published by news outlets and social media, which are increasingly being used assources for any direct updates by investors. In this paper, an approach was presented to solve the problem of Stock Price Prediction. In the first chapter, an overview of the context was given, starting from the early years when the first methodologies were used to sup-port investors in their transactions, to the last years when artificial intelligence has been adopted for the resolution of numerous problems in the financial sector. The second chapter introduces state-of-the-art articles which show the architectures that have been designed and the innovations to solve the task related to the prediction of market trends. The third chapter illustrated the architecture used in this thesis in order to solve the Stock Price Prediction task, introducing sentiment techniques analysis in order to be able to assign a value to sentiment derived from news published online and comments on social media. The fourth chapter introduces the research questions that this paper aims to answer from the results obtained. Finally, the fifth chapter shows the results and explains the reasoning that made possible to answer the research questions.