Time: 1245pm-145pm, Friday, March 5, 2010
Place: Room 6496, CUNY Graduate Center, 365 Fifth Ave (34str&35str).
Speaker: Lijun Feng (CUNY)
Title: Automatic Readability Assessment

Abstract:

What makes a text easy or hard to read? Previous research in automatic
readability assessment has looked at a limited class of lexical and
syntactic properties of texts. In this study, we explore linguistic
features at several levels (lexical, syntactic, discourse, etc.) to
investigate and analyze their possible impact on and correlations with
text readability.

In addition to systematically analyzing well studied lexical and
syntactic features and expanding new features at these levels, we
investigate a set of novel features at various discourse levels which
may impact text readability from a text comprehension point of view. We
combine NLP and machine learning techniques to build and evaluate an
automatic readability assessment tool based on these features. Our best
performing model derived from stepwise greedy approach showed
significant improvement compared with previous studies in the field.

Speaker Bio:

Lijun Feng is a PhD student in the Computer Science Department at the
Graduate Center,  the City University of New York (CUNY). Her research
interests include natural language processing (NLP), in particular
readability, text comprehension and text simplification, and machine
learning.  Her thesis research combines NLP and machine learning
techniques to build and evaluate an automatic text readability
assessment tool, both for general audience and readers with intellectual
disabilities in particular.