Lucas Carstens

Imperial College London

Research

Abstract

I am currently developing approaches to Argument Mining that combine statistical methods with rule-based ones. Argument Mining is concerned with the (automated) construction of Argument Frameworks from text. Argument Frameworks are comprised of arguments, i.e. premises and related conclusions, and connections between such arguments, forming a network, or framework, of textual entities and relations between them. Other research topics I am interested in include

  • Sentiment Analysis
  • Computational Linguistics
  • Cognitive Science
  • Human Computer Interaction (HCI)

Projects

Automating relation classification in an online debating platform

quaestio-it is an online social debating platform that applies Argumentation theory to optimise online debate. We aim to provide facilities to autmatically determine relations between arguments posted by users

DesMold

Funded by the 7th Framework Programme, DesMOLD aims at creating a new generation of knowledge-based engineering systems for added-value injection molding and process control optimization

Enhancing Sentiment Extraction from text by means of Arguments

During my MSc project I developed a prototype that improves traditional Sentiment Classification (using Support Vector Machines) by adding user generated arguments to gain additional insights into the sentiment of pieces of text. This project was recognised as one of five distinguished projects in the class of 2011

Vince - Theory driven design and Implementation of an Embodied Conversational Agent

During the final year project of my undergraduate studies I have implemented an Embodied Conversational agent that acts as a personal health assistant in an office environment

Experience

PhD candidate

Imperial College London October 2012 - present
During my PhD I am working at the crossroads of Machine Learning, Argumentation and Linguistics. My main focus lies on building Argument Frameworks, networks of arguments and relations between them, from free text. I am a member of the Computational Logic and Argumentation (CLArg) group and the Statistical Machine Learning group under the supervision of Professor Francesca Toni and Dr. Marc Deisenroth

Research Assistant

Imperial College London October 2012 - September 2015
Funded by the 7th Framework Programme, DesMOLD aims at creating a new generation of knowledge-based engineering systems for added-value injection molding and process control optimization. Our group at Imperial is responsible for the development of an Argumentation module that facilitates the decision making process based on past cases and knowlegde of stake holders

Summer Associate

Thomson Reuters June 2014 - August 2014
Working with the Research & Development team in London, UK, and Eagan (MN), USA, on automated text summarisation

Technical Analyst

Everis October 2011 - September 2012
Working for this Consultancy I was involved in a range of projects, most notably working on one of Europe's largest data integration & migration projects. This included supervising changes to and merges of core and legacy systems from two separate Retail Banking platforms.

MSc Computing, Aritifical Intelligence

Imperial College London September 2010 - September 2011

BSc Applied Cognitive Science & Media Science

University of Duisburg-Essen, Germany October 2007 - August 2010

Research Assistant & tutor

University of Duisburg-Essen, Germany September 2009 - August 2010

Publications

2015

Lucas Carstens, Francesca Toni: Improving out-of-domain Sentiment Polarity Classification using Argumentation, SENTIRE 2015

Lucas Carstens, Francesca Toni, Valentinos Evripidou: Towards Relation-based Argumentation Mining, 2nd Workshop on Argumentation Mining

Lucas Carstens, Xiuyi Fan, Yang Gao, Francesca Toni: An overview of Argumentation Frameworks for Decision Support, GKR 2015

2014

Lucas Carstens, Francesca Toni, Valentinos Evripidou: Argumentation Mining and Social debates, COMMA 2014

Valentinos Evripidou, Lucas Carstens, Francesca Toni, David Cabanillas: Argumentation-Based Collaborative Decisions for Design, ICTAI 2014

2013

Lucas Carstens, Francesca Toni: Enhancing sentiment extraction from text by means of arguments. WISDOM 2013

David Cabanillas, Francesc Bonada, Raquel Ventura, Francesca Toni, Valentinos Evripidou, Lucas Carstens, Liceth Rebolledo: A Combination of Knowledge and Argumentation Based System for Supporting Injection Mould Design. CCIA 2013

2010

Sabrina Eimler, Astrid von der Pütten, Ulrich Schächtle, Lucas Carstens, Nicole Krämer: Following the White Rabbit – A Robot Rabbit as Vocabulary Trainer for Beginners of English. HCI in Work and Learning, Life and Leisure. Lecture Notes in Computer Science Volume 6389, 2010, pp 322-339

Contact

Drop me a line at

lucas dot carstens10 at imperial dot ac dot uk

Contact me via LinkedIn

Downloads

Argumentation Mining corpus