Multi-Channel CNN Based Inner-Attention for Compound Sentence Relation Classification

Kaili Sun,Yuan Li,Dunhua Deng,Yang Li

doi:10.1109/access.2019.2943545

Kaili Sun, Yuan Li + Show 2 more

Open Access

PDF Available

https://doi.org/10.1109/access.2019.2943545

Copy DOI

Export

Save

Cite

Journal: IEEE Access	Publication Date: Jan 1, 2019
Citations: 14	License type: CC BY 4.0

Affiliation: Central China Normal University

Abstract
Highlights/Summary
Full-Text PDF
Similar Papers

Abstract

Listen

Relation classification is a vital task in natural language processing, and it is screening for semantic relation between clauses in texts. This paper describes a study of relation classification on Chinese compound sentences without connectives. There exists an implicit relation in a compound sentence without connectives, which makes it difficult to realize the recognition of relation. The major challenges that relation classification modeling faces are how to obtain the contextual representation of sentence and relation dependence features between clauses. To solve this problem, we propose a novel Inatt-MCNN model to extract sentence features and classify relations by combining multi-channel CNN and Inner-attention mechanism. This network structure utilizes CNN to extract local features of sentences and Inner-attention to capture sentence-level feature representations for this relation classification task. Besides, since the Inner-attention is based on Bi-LSTM, the global and long-term dependence semantic information can be well obtained in Inatt-MCNN to promote the model performance. We conduct experiments on two public Chinese discourse datasets: the Chinese compound sentence corpus (CCCS) dataset and the Tsinghua Chinese Treebank(TCT) dataset. Compared with the previous public methods, Inatt-MCNN model has superior performance and achieves the highest accuracy, especially on the CCCS dataset.

Highlights

Besides character, word, and phrase, sentence is an important level of research in natural language processing(NLP) applications
We propose a joint model Inatt-MCNN, which is a combination of the Inner-attention mechanism and multi-channel Convolutional neural network(CNN), to solve the sentence representation and relation classification task
RELATIONSHIP CLASSIFICATION DATASETS The performance of the proposed model is evaluated on two public Chinese discourse datasets: the Chinese compound sentence corpus(CCCS) [36] and the Tsinghua Chinese Treebank(TCT) [37], which are presently the datasets that mainly contain Chinese compound sentences

Summary

Introduction

Word, and phrase, sentence is an important level of research in natural language processing(NLP) applications. Compound sentence relation classification is an indispensable part of compound sentence research, and it is a basic research problem in the understanding of natural language. A. CONVOLUTIONAL NEURAL NETWORKS The CNN is widely used in image processing [20], target detection [21], and even medical discovery [22]. CNN can have numerous convolutional layers, which contains nonlinear activation functions such as tanh or ReLu. And different from classical feedforward neural network, CNN in each layer can use different size kernels, which have hundreds or thousands of filters, over the input layer to compute the output. The pooling operation [25] in the pooling layer, which is applied to the feature map of a convolutional layer. The commonly used pooling operations include average pooling, max Pooling, and stochastic pooling

Methods

Results

Conclusion