Learning Distributed Representations of Phrases
Konstantin Lopyrev · 2014
Recent work in Natural Language Processing has focused on learning distributed representations of words, phrases, sentences, paragraphs and even whole documents. In such representations, text is represented using multi-dimensional vectors and similarity between pieces of text can be measured using similarity between such vectors. In this project I focus my attention on learning representations of phrases - sequences of two or more words that can function as a single unit in a sentence.