Recursive Deep Models for Semantic Compositionality

me@ryanorban.com (Ryan Orban) — Thu, 05 Sep 2013 00:00:00 +0000

Summary

Richard Socher and colleagues at Stanford NLP published this work introducing the Stanford Sentiment Treebank and a Recursive Neural Tensor Network (RNTN) that achieved state-of-the-art sentiment analysis results by operating on constituency parse trees. Rather than treating a sentence as a bag of words or a flat sequence, the RNTN computed sentiment at every internal node of a parse tree — capturing how negation, intensifiers, and compositional structure change meaning.

Recursive-Neural-Networks on Ryan Orban

Recursive Deep Models for Semantic Compositionality

Summary