Current research in automatic single document summarization is dominated by
two effective, yet naive approaches: summarization by sentence extraction, and
headline generation via bag-of-words models. While successful in some tasks,
neither of these models is able to adequately capture