The State of the Art in Semantic Parsing

Speaker:
Percy Liang
Abstract:
Semantic parsing, the task of mapping utterances to semantic  representations (e.g. logical forms), has its roots in the early natural language understanding systems of the 1960s. These rule-based systems were representationally sophisticated, but brittle, and thus fell out of favor as the statistical revolution swept NLP. Since the late 1990s, however, there has been a resurgence of interest in  semantic parsing from the statistical perspective, where the representations are logical but the learning is not. Most recently, there are efforts to learn these logical forms automatically from denotations, a much more realistic but also challenging setting. The learning perspective has both led to practical large-scale semantic parsers, but interestingly also has implications for the semantic representations.
Length:
01:11:24
Date:
10/07/2014
views: 1100

Images:
Preview of 10b1.jpg
Image 10b1.jpg
Preview of 10b2.jpg
Image 10b2.jpg
Preview of 10b3.jpg
Image 10b3.jpg
Attachments: (video, slides, etc.)
66 MB
228 downloads
440 MB
321 downloads
47 MB
303 downloads