Abstract
Text Classification is an important and classical problem in natural languageprocessing. There have been a number of studies that applied convolutionalneural networks (convolution on regular grid, e.g., sequence) toclassification. However, only a limited number of studies have explored themore flexible graph convolutional neural networks (convolution on non-grid,e.g., arbitrary graph) for the task. In this work, we propose to use graphconvolutional networks for text classification. We build a single text graphfor a corpus based on word co-occurrence and document word relations, thenlearn a Text Graph Convolutional Network (Text GCN) for the corpus. Our TextGCN is initialized with one-hot representation for word and document, it thenjointly learns the embeddings for both words and documents, as supervised bythe known class labels for documents. Our experimental results on multiplebenchmark datasets demonstrate that a vanilla Text GCN without any externalword embeddings or knowledge outperforms state-of-the-art methods for textclassification. On the other hand, Text GCN also learns predictive word anddocument embeddings. In addition, experimental results show that theimprovement of Text GCN over state-of-the-art comparison methods become moreprominent as we lower the percentage of training data, suggesting therobustness of Text GCN to less training data in text classification.