Hooks in the Headline: Learning to Generate Headlines with Controlled Styles

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 5082–5093 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 5082 Hooks in the Headline: Learning to Generate Headlines with Controlled Styles Di Jin,1 Zhijing Jin,2 Joey Tianyi Zhou,3∗Lisa Orii,4 Peter Szolovits1 1CSAIL, MIT, 2Amazon Web Services, 3A*STAR, Singapore, 4Wellesley College {jindi15,psz}@mit.edu, zhijing.jin@connect.hku.hk zhouty@ihpc.a-star.edu.sg, lorii@wellesley.edu Abstract Current summarization systems only produce plain, factual headlines, but do not meet the practical needs of creating memorable titles to increase exposure. We propose a new task, Stylistic Headline Generation (SHG), to enrich the headlines with three style options (humor, romance and clickbait), in order to attract more readers. With no style-specific article-headline pair (only a standard headline summarization dataset and mono-style corpora), our method TitleStylist generates style-specific headlines by combining the summarization and recon- struction tasks into a multitasking framework. We also introduced a novel parameter sharing scheme to further disentangle the style from the text. Through both automatic and human evaluation, we demonstrate that TitleStylist can generate relevant, fluent headlines with three target styles: humor, romance, and click- bait. The attraction score of our model gen- erated headlines surpasses that of the state-of- the-art summarization model by 9.68%, and even outperforms human-written references.1 1 Introduction Every good article needs a good title, which should not only be able to condense the core meaning of the text, but also sound appealing to the read- ers for more exposure and memorableness. How- ever, currently even the best Headline Generation (HG) system can only fulfill the above requirement yet performs poorly on the latter. For example, in Figure 1, the plain headline by an HG model “Summ: Leopard Frog Found in New York City” is less eye-catching than the style-carrying ones such as “What’s That Chuckle You Hear? It May Be the New Frog From NYC.” ∗Corresponding author. 1Our code is available at https://github.com/ jind11/TitleStylist. New frog species discovered in New York City area. It has a distinctive croak, scientists find. Leopard frog species doesn't yet have a name. Ribbit! Frog Species Found in New York City Has a Croak of Its Own Original Headline: Article Summ: Leopard Frog Found in New York City HG Model Output: What's that Chuckle You Hear? It May be the New Frog from NYC Humorous: A New Frog with a Croak of Its Own Awaits its Name in the Roads of NYC Romantic: 3 Facts about the New Frog with a Croak of Its Own Click-Baity: Figure 1: Given a news article, current HG models can only generate plain, factual headlines, failing to learn from the original human reference. It is also much less attractive than the headlines with humorous, romantic and click-baity styles. To bridge the gap between the practical needs for attractive headlines and the plain HG by the current summarization systems, we propose a new task of Stylistic Headline Generation (SHG). Given an article, it aims to generate a headline with a target style such as humorous, romantic, and click-baity. It has …

Content truncated to 500 words