Frank Xing, PhD




                 

                 

Bio

※ I spent my boyhood in Wuhan, a punk metropolis in central China. But illumination seldom comes to my rescue and all the time seems lost in history before I moved to Beijing in 2011. However, my enthusiasm for exploring life in the ancient capital faded after I received my bachelor's degrees in Information Systems and Economics from Peking University. So I drifted to Singapore and did my PhD at Nanyang Technological University, co-advised by Prof. Erik Cambria and Prof. Roy Welsch. After the one-year industrial experience with AIR Lab at Continental, I am Presidential Postdoctoral Fellow at Nanyang Technological University today, working with Prof. Sun Aixin and Prof. Erik Cambria.

※ My research interest spans NLP, Sentiment Analysis, Business Intelligence, and AI empowered finance.

※ Academic genealogy, a.k.a. how you do research [expand].

※ My Erdös number is 3 and Markowitz number is 4.

News

※ I am actively looking for a faculty position.

※ New book "Intelligent Asset Management" available from Amazon.com!

Projects

Only active projects and their relevant publications are described here. For a full list of publications see DBLP.

AI and Financial Markets

The fast spread of information across news and social media have made financial markets increasingly volatile. Traditional models in finance, developed decades ago, has overlooked public opinions and sentiment. This project leverages on sentic computing and other NLP methods to monitor public moods and discussions. This new information is further integrated to the stock/forex price prediction and asset allocation models to help improve them.

Asset Dependence Modeling

Estimating correlations among different assets is an indispensable step under the the framework of modern portfolio theory. While the correlations induced from price data are unreliable, especially in high-dimensional space. This project seeks alternative sources of information using NLP techniques, to facilitate robust and large-scale modeling of the dependence structure. The methods developed have great applicational potential for portfolio risk management.

Financial Knowledge Engineering

Language in financial domain is characterized by its frequent use of jargons and commonsense knowledge. Numbers, time expressions, and arguments, which are trivial in general domain without context play an important role in understanding financial texts. This project endeavors to combine the top-down ontology construction approach and bottom-up machine learning techniques to build a knowledge base for financial analysis and reasoning.