Recursive Deep Models for Semantic Compositionality Over a Sentiment Treebank

TLDR

Semantic word spaces are useful but cannot capture the meaning of longer phrases, and advancing sentiment compositionality requires richer training data and more powerful composition models. The authors introduce a Sentiment Treebank and a Recursive Neural Tensor Network to address these challenges. The Sentiment Treebank provides fine‑grained sentiment labels for 215,154 phrases in 11,855 parse trees, and the Recursive Neural Tensor Network learns to compose phrase meanings over these trees. Trained on the treebank, the model surpasses prior methods, raising single‑sentence sentiment accuracy from 80 % to 85.4 %, achieving 80.7 % fine‑grained phrase accuracy—an 9.7 % gain over bag‑of‑features baselines—and uniquely captures negation effects at multiple tree levels.

Abstract

Semantic word spaces have been very useful but cannot express the meaning of longer phrases in a principled way. Further progress towards understanding compositionality in tasks such as sentiment detection requires richer supervised training and evaluation resources and more powerful models of composition. To remedy this, we introduce a Sentiment Treebank. It includes fine grained sentiment labels for 215,154 phrases in the parse trees of 11,855 sentences and presents new challenges for sentiment compositionality. To address them, we introduce the Recursive Neural Tensor Network. When trained on the new treebank, this model outperforms all previous methods on several metrics. It pushes the state of the art in single sentence positive/negative classification from 80% up to 85.4%. The accuracy of predicting fine-grained sentiment labels for all phrases reaches 80.7%, an improvement of 9.7% over bag of features baselines. Lastly, it is the only model that can accurately capture the effects of negation and its scope at various tree levels for both positive and negative phrases.