Liangyou Li - Transfer Talk - 19th June 2015

Video Category: 
Transfer Talk
Liangyou Li

Title: Graph-Based Machine Translation

Supervisor: Prof. Qun Liu, Prof. Andy Way

Abstract:

Compared to sequences and trees, which have dominated the field of natural language processing (NLP) for decades, graphs are more general and powerful for modelling lan- guages. In both computational and theoretical linguistics, graphs are used to represent many morpho-syntactic aspects of languages. However, because of their high complexity, graphs are not widely used in NLP. Recently, hypergraphs, which is used to represent the meaning of a sentence, have been revisited and used for semantic-based machine translation (MT). This trial and recent progress on neural network MT encourage us to explore graph-based MT to make use of various type of linguistic knowledge in a unified model.