Paper: Automatic Metrics for Genre-specific Text Quality

ACL ID N12-2010
Title Automatic Metrics for Genre-specific Text Quality
Venue Annual Conference of the North American Chapter of the Association for Computational Linguistics
Session Student Session
Year 2012

To date, researchers have proposed different ways to compute the readability and coher- ence of a text using a variety of lexical, syn- tax, entity and discourse properties. But these metrics have not been defined with special rel- evance to any particular genre but rather pro- posed as general indicators of writing qual- ity. In this thesis, we propose and evalu- ate novel text quality metrics that utilize the unique properties of different genres. We fo- cus on three genres: academic publications, news articles about science, and machine gen- erated text, in particular the output from auto- matic text summarization systems.