Improving Semantic Relevance for Sequence-to-Sequence Learning of Chinese Social Media Text Summarization

TL;DR

This paper proposes a neural model that enhances semantic relevance in Chinese social media text summarization by maximizing similarity between source and summary representations, leading to improved performance.

Contribution

It introduces a semantic relevance-based neural model with a gated attention encoder and similarity maximization during training for better summarization quality.

Findings

Outperforms baseline systems on social media corpus

Achieves higher semantic relevance in summaries

Demonstrates effectiveness of similarity maximization approach

Abstract

Current Chinese social media text summarization models are based on an encoder-decoder framework. Although its generated summaries are similar to source texts literally, they have low semantic relevance. In this work, our goal is to improve semantic relevance between source texts and summaries for Chinese social media summarization. We introduce a Semantic Relevance Based neural model to encourage high semantic similarity between texts and summaries. In our model, the source text is represented by a gated attention encoder, while the summary representation is produced by a decoder. Besides, the similarity score between the representations is maximized during training. Our experiments show that the proposed model outperforms baseline systems on a social media corpus.