Loading
COGNITION & LANGUAGE
Andrew Maas: Learning Word Representations for Semantics and Sentiment
Thursday October 11, 2012 | 03:00 -04:00 PM | Cordura Hall conference room

Unsupervised vector-based approaches to semantics can model rich lexical meanings, but they largely fail to capture sentiment information that is central to many word meanings and important for a wide range of NLP tasks. We present a model that uses a mix of unsupervised and supervised techniques to learn word vectors capturing semantic term–document information as well as rich sentiment content. The proposed model can leverage both continuous and multi-dimensional sentiment information as well as non-sentiment annotations. We demonstrate the model can capture binary notions of sentiment in a movie review classification task. Further, we evaluate the model using data from the Experience Project -- online forum texts with rich multi-dimensional sentiment annotations. Beyond our specific models, we believe multi-dimensional notions of emotive expression and using sentiment annotations to build word representations are interesting topics for further research in artificial intelligence and pragmatics.