Category Archives: reading

Sentiment analysis in a character-level model

Learning to Generate Reviews and Discovering Sentiment. Alec Radford, Rafal Jozefowicz, Ilya Sutskever. https://arxiv.org/abs/1704.01444

In this paper (apparently only published thru arxiv, so not carefully reviewed by anyone just yet) the authors present an intriguing result. They build a neural-inspired model (LSTM, a fairly standard one) which predicts the next byte in a text, given the ones it already has seen. They train the model on product reviews, and then use it as an input to a simple classifier. The model, in spite of being trained on characters, does very well (better than many standard lexical models, e.g.) on classifying sentiment of product reviews! The authors even find (to their own delight) an indicator cell specifically for sentiment, and show how it tracks sentiment along the progression of the text. This may seem strange, but actually there is a fairly reasonable hypothesis to explain the result: there is more to sentiment than lexical resources can model. This model appears to capture signal which is encoded in something more than the sequence of words.

In general, coercing most everything about language into lexical models (as recent results have done) is fixing the representation on one analysis level which happens to be accessible due to the nature of our writing system. Breaking this strong binding is probably a good idea.

Associative meaning

Deese, James. 1962. On the structure of associative meaning. Psychological Review, Vol 69(3), 161-175. http://dx.doi.org/10.1037/h0045842

Deese, who has published extensively on association norms and the methodology of eliciting associations, discusses here what sort of relation the terms in associative pairs might have. Most of the paper notes that the methodology heretofore has been faulty, and Deese’s contribution is to introduce frequency well into the model. He also discusses associative relations in terms of replaceability and combinability and in the assymmetry between items on different hyponymi levels.

Deese posits (referring to previous work by Woodworth, Ebbinghaus, and Galton, to which I expect I might return further on) that the associative relation between elicited terms is not one of meaning in the way meaning usually is understood. (Woodworth per Deese, classifies (grades, probably) words both by meaning and meaningfulness. This needs to be looked up properly). The associative relation is not as readily mappable on known lexicogrammatic relations.