Graph Convolutional Networks for Text Classification

Abstract

Text classification is an important and classical problem in natural language processing. There have been a number of studies that applied convolutional neural networks (convolution on regular grid, e.g., sequence) to classification. However, only a limited number of studies have explored the more flexible graph convolutional neural networks (convolution on non-grid, e.g., arbitrary graph) for the task. In this work, we propose to use graph convolutional networks for text classification. We build a single text graph for a corpus based on word co-occurrence and document word relations, then learn a Text Graph Convolutional Network (Text GCN) for the corpus. Our Text GCN is initialized with one-hot representation for word and document, it then jointly learns the embeddings for both words and documents, as supervised by the known class labels for documents. Our experimental results on multiple benchmark datasets demonstrate that a vanilla Text GCN without any external word embeddings or knowledge outperforms state-of-the-art methods for text classification. On the other hand, Text GCN also learns predictive word and document embeddings. In addition, experimental results show that the improvement of Text GCN over state-of-the-art comparison methods become more prominent as we lower the percentage of training data, suggesting the robustness of Text GCN to less training data in text classification.

An Ontology-Aware Framework for Audio Event Classification

“…Graph Convolutional Networks The second component of our solution is Graph Convolutional Networks (GCNs). Recently, GCNs have achieved immense success in capturing the underlying associations and correlations between entities and have been widely adapted to various domains such as computer vision [7] and natural language processing [12,13]. However, to the best of our knowledge, GCNs have never been explored in audio domain.…”

Dual-Attention Graph Convolutional Network

“…Some other methods like [27] regard documents or sentences as the graphs of words. Differently, Yao et.al [26] propose a new way to construct the graph by regarding both documents and words as nodes, which performs quite well with GCN.…”

“…The whole architecture of the proposed dual-attention GCN framework is shown in Fig.1, where the input is a graph based on a given text. For graph construction, we adopt the method proposed in [26] which removes useless words in texts first and then models both the text and its words as nodes. This process is described in detail in Section 3.2.…”

“…In experiments, we use pre-trained embedding features from TextGCN [26] with the size of 200. In the process of constructing the graph, we set the PMI window size as 20 to be more comparable.…”

Edge-based sequential graph generation with recurrent neural networks

“…This trend has been increasing especially since the advent of Graph Neural Networks [4] and contextual Neural Networks for Graphs [5], which paved the road for modern graph-based Deep Learning [6] models. As of today, Graph Neural Networks are used with success for predictive tasks such as semi-supervised classification [7], link prediction [8], and text classification [9].…”