Text Coherence Analysis Based on Deep Neural Network

TL;DR

This paper introduces a deep neural network model that effectively assesses text coherence by learning sentence representations and their interactions, demonstrating significant improvements in sentence ordering tasks.

Contribution

It presents a novel deep coherence model using CNNs that jointly learns sentence representations and coherence, trained end-to-end for the first time.

Findings

Significant improvement over state-of-the-art in sentence ordering

Effective end-to-end training of coherence model

Demonstrated robustness in coherence assessment

Abstract

In this paper, we propose a novel deep coherence model (DCM) using a convolutional neural network architecture to capture the text coherence. The text coherence problem is investigated with a new perspective of learning sentence distributional representation and text coherence modeling simultaneously. In particular, the model captures the interactions between sentences by computing the similarities of their distributional representations. Further, it can be easily trained in an end-to-end fashion. The proposed model is evaluated on a standard Sentence Ordering task. The experimental results demonstrate its effectiveness and promise in coherence assessment showing a significant improvement over the state-of-the-art by a wide margin.