Supervised and Semi-Supervised Text Categorization using LSTM for Region Embeddings

TL;DR

This paper introduces a novel approach using LSTM-based region embeddings for text categorization, outperforming previous methods by effectively capturing complex text concepts in both supervised and semi-supervised settings.

Contribution

It proposes a new LSTM-based region embedding method for text categorization, demonstrating superior performance over existing models on benchmark datasets.

Findings

LSTM region embeddings outperform CNN-based embeddings.

Combining LSTM and convolution layers improves accuracy.

The approach exceeds previous best results on four datasets.

Abstract

One-hot CNN (convolutional neural network) has been shown to be effective for text categorization (Johnson & Zhang, 2015). We view it as a special case of a general framework which jointly trains a linear model with a non-linear feature generator consisting of `text region embedding + pooling'. Under this framework, we explore a more sophisticated region embedding method using Long Short-Term Memory (LSTM). LSTM can embed text regions of variable (and possibly large) sizes, whereas the region size needs to be fixed in a CNN. We seek effective and efficient use of LSTM for this purpose in the supervised and semi-supervised settings. The best results were obtained by combining region embeddings in the form of LSTM and convolution layers trained on unlabeled data. The results indicate that on this task, embeddings of text regions, which can convey complex concepts, are more useful than embeddings of single words in isolation. We report performances exceeding the previous best results on four benchmark datasets.