The Style-Content Duality of Attractiveness: Learning to Write Eye-Catching Headlines via Disentanglement

Abstract

Eye-catching headlines function as the first device to trigger more clicks, bringing reciprocal effect between producers and viewers. Producers can obtain more traffic and profits, and readers can have access to outstanding articles. When generating attractive headlines, it is important to not only capture the attractive content but also follow an eye-catching written style. In this paper, we propose a Disentanglement-based Attractive Headline Generator (DAHG) that generates headline which captures the attractive content following the attractive style. Concretely, we first devise a disentanglement module to divide the style and content of an attractive prototype headline into latent spaces, with two auxiliary constraints to ensure the two spaces are indeed disentangled. The latent content information is then used to further polish the document representation and help capture the salient part. Finally, the generator takes the polished document as input to generate headline under the guidance of the attractive style. Extensive experiments on the public Kuaibao dataset show that DAHG achieves state-of-the-art performance. Human evaluation also demonstrates that DAHG triggers 22% more clicks than existing models.

arXiv commentAccepted by AAAI2021

References in corpus (17)

• Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation
• InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets
• SummaRuNNer: A Recurrent Neural Network based Sequence Model for Extractive Summarization of Documents
• A Hierarchical Neural Autoencoder for Paragraphs and Documents
• Learning Discourse-level Diversity for Neural Dialog Models using Conditional Variational Autoencoders
• Get To The Point: Summarization with Pointer-Generator Networks
• A Unified Model for Extractive and Abstractive Summarization using Inconsistency Loss
• Neural Document Summarization by Jointly Learning to Score and Select Sentences
• Automatic Article Commenting: the Task and Dataset
• Neural Latent Extractive Document Summarization
• VMSMO: Learning to Generate Multimodal Summary for Video-based News Articles
• Hooks in the Headline: Learning to Generate Headlines with Controlled Styles
• From Standard Summarization to New Tasks and Beyond: Summarization with Manifold Information
• Infusing Sequential Information into Conditional Masked Translation Model with Self-Review Mechanism
• Reasoning in Dialog: Improving Response Generation by Context Reading Comprehension
• Iterative Document Representation Learning Towards Summarization with Polishing
• Clickbait? Sensational Headline Generation with Auto-tuned Reinforcement Learning