Any-gram Kernels for Sentence Classification: A Sentiment Analysis Case Study

TL;DR

This paper introduces an improved and more efficient version of any-gram kernels for sentence classification, particularly sentiment analysis, that effectively incorporates word embeddings and outperforms previous methods.

Contribution

The authors develop a tree-kernel-independent, more efficient any-gram kernel approach and enhance the use of word embeddings for sentiment classification.

Findings

Significantly improved sentiment classification accuracy

Efficient kernel computation without tree structures

Effective integration of word embeddings

Abstract

Any-gram kernels are a flexible and efficient way to employ bag-of-n-gram features when learning from textual data. They are also compatible with the use of word embeddings so that word similarities can be accounted for. While the original any-gram kernels are implemented on top of tree kernels, we propose a new approach which is independent of tree kernels and is more efficient. We also propose a more effective way to make use of word embeddings than the original any-gram formulation. When applied to the task of sentiment classification, our new formulation achieves significantly better performance.