On the Learnability of Concepts: With Applications to Comparing Word Embedding Algorithms
This work addresses the need for a systematic comparison of word embedding algorithms for NLP researchers, but it is incremental as it builds on existing methods with a new analysis framework.
The paper tackles the problem of comparing word embedding algorithms by introducing a notion of 'concept' as a list of semantically related words and measuring learnability via classifier performance on unseen members. It finds that all embedding methods capture semantic content, with fastText performing better than others.
Word Embeddings are used widely in multiple Natural Language Processing (NLP) applications. They are coordinates associated with each word in a dictionary, inferred from statistical properties of these words in a large corpus. In this paper we introduce the notion of "concept" as a list of words that have shared semantic content. We use this notion to analyse the learnability of certain concepts, defined as the capability of a classifier to recognise unseen members of a concept after training on a random subset of it. We first use this method to measure the learnability of concepts on pretrained word embeddings. We then develop a statistical analysis of concept learnability, based on hypothesis testing and ROC curves, in order to compare the relative merits of various embedding algorithms using a fixed corpora and hyper parameters. We find that all embedding methods capture the semantic content of those word lists, but fastText performs better than the others.