Improving Grammaticality in Statistical Sentence Generation

Add to your list(s) Download to your calendar using vCal

• Yue Zhang
• Monday 24 May 2010, 12:30-13:30
• GS15, Computer Laboratory.

If you have a question about this talk, please contact Diarmuid Ó Séaghdha.

At this session of the NLIP Reading Group we’ll be discussing the following paper:

Stephen Wan, Mark Dras, Robert Dale and Cécile Paris. 2009. Improving Grammaticality in Statistical Sentence Generation: Introducing a Dependency Spanning Tree Algorithm with an Argument Satisfaction Model. Proceedings of EACL -09.

Abstract: Abstract-like text summarisation requires a means of producing novel summary sentences. In order to improve the grammaticality of the generated sentence, we model a global (sentence) level syntactic structure. We couch statistical sentence generation as a spanning tree problem in order to search for the best dependency tree spanning a set of chosen words. We also introduce a new search algorithm for this task that models argument satisfaction to improve the linguistic validity of the generated tree. We treat the allocation of modifiers to heads as a weighted bipartite graph matching (or assignment) problem, a well studied problem in graph theory. Using BLEU to measure performance on a string regeneration task, we found an improvement, illustrating the benefit of the spanning tree approach armed with an argument satisfaction model.

This talk is part of the Natural Language Processing Reading Group series.

This talk is included in these lists:

• Cambridge Forum of Science and Humanities
• Cambridge Language Sciences
• Cambridge talks
• Chris Davis' list
• GS15, Computer Laboratory
• Guy Emerson's list
• Natural Language Processing Reading Group

Note that ex-directory lists are not shown.