Marilisa Amoia

Marilisa Amoia
research category image

Main area of research
Natural language & AI

Marilisa Amoia’s research interest is in the area of Natural Language Processing. After the Master's program in Computational Linguistics at the Saarland University in 2003, she started her PhD that was jointly supervised by the Saarland University and the University of Nancy, France. During the PhD years, she has worked on linguistic-based approaches to paraphrase recognition and inference in natural language. At that time, she was a researcher at Loria in Nancy in the Language et Dialogue Group. In November 2008, she defended her PhD thesis “Linguistic Based Computational Treatment of Textual Entailment Recognition”. In 2009, she joined the TALARIS Lab at INRIA in Nancy. Here, her work focused on real-virtual world interactions within a virtual travel scenario and in particular on generating natural language, easy-to-understand navigational instructions enabling user to navigate in 3D-virtual worlds. She also worked on the design and implementation of an interactive 3D serious game for second language acquisition integrated in Second Life. From May 2011 till September 2013, she worked at the Applied Linguistic Department at the University of Saarland on several different projects. In the GECCO project, her tasks mainly included corpus management and application of quantitative methods for information retrieval. In the CLARIN-D project, she worked on the design of enhanced user-friendly interfaces allowing non-technical experts classifying and retrieving knowledge from databases in the field of e-humanities. She contributed to the TELL project by designing and implementing a learning platform for second language acquisition in the medical domain. She also taught at the University of Saarland for many years Prolog programming courses and introductory courses on statistical NLP techniques in the language R. In october 2013, Marilisa joined the NLU language team Italian at Nuance.

Selected articles

Using Set Constraints to generate Distinguishing Descriptions
Generation of definite descriptions with set constraints
Generating Definite Descriptions: Non incrementality, inference and data
Paraphrastic grammars
Recognition of Alternation Paraphrases: a robust and exhaustive approach

1 2 3 4