PhD position: Deep Learning and Random Forest for Argumentation Mining

Posted by: Michael SCHYNS
Date:2015-09-22
Contact:[email protected]
Attachmentone attached document

PhD Position: Deep Learning and Random Forest for Argumentation Mining

The research group QuantOM (Center for Quantitative Methods and Operations Management), Operations Department, of HEC-Management School of the University of Liège (Belgium) has a vacancy for one full-time Ph. D. scholarship for four years starting in the fall/winter 2015 or beginning of 2016.

The main aim of the PhD project will be to develop novel machine learning techniques for argumentation mining (AM). AM is a sub-field of Natural Language Processing, and involves developing algorithms for automatically detecting argumentative structures (e.g. support, conclusion, counter-argument) from text data (e.g. discourse). Specifically, the selected candidate will be expected to design, implement and evaluate innovative AM techniques based on Deep Learning and Random Forest, which are two recent and promising machine learning paradigms. In addition to these, the candidate will also be expected to investigate (and evaluate) other state-of-the-art machine learning approaches for AM, in particular, minimally-supervised learning and distant supervision, as well as more classical approaches, such as support vector machines. The machine learning techniques for AM developed in this project will constitute the first contribution of this research. In the next step, the candidate will investigate whether argumentative structures, discovered by the proposed algorithms from online reviews (e.g. Amazon, YELP, TripAdvisor), are predictive of the reviews’ helpfulness score. Establishing such correlations will be performed using regressions and time-series analysis, and constitutes the second research contribution.

We are looking for a candidate with a very strong interest in machine learning research. The candidate should be proficient in programming (e.g. R, Matlab, Java, Python) and in quantitative methods (e.g. regressions, time-series, vector algebra, and statistics), and hold a Masters degree in Computer Science, Operations Research or Applied Mathematics.

Interested candidates are invited to read the accompanying document for more information on the PhD project, the position itself and the application procedure.

Contact person and PhD supervisor: Ashwin Ittoo, [email protected]