Linyi Yang

The explosion in the sheer magnitude and complexity of textual data in recent years makes it increasingly challenging for investment analysts to extract valuable insights. While progress has been made in applying machine learning, text mining and semantic analysis to tackle challenges, current solutions are still far from taking full advantages of the rich information on the Internet. In general, existing methodologies are incapable of connecting various data sources in an accurate way for financial forecast tasks, while the finance researchers typically perform analysis on textual data at a superficial level. Therefore, the proposed research project aims to connect unstructured data from multiple sources (e.g. news, social media, financial report) to extract interpretable relationships between entities, predict the asset prices, and quantify financial risk and uncertainty. Building on recent advances in Natural Language Processing, semantic analysis, and deep neural networks, the project demands not only the ability to quantify and forecast the probabilistic in the financial market, but also explain it with textual information.