Diverse, Controllable, and Keyphrase-Aware: A Corpus and Method for News Multi-Headline Generation

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 6241–6250, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 6241 Diverse, Controllable, and Keyphrase-Aware: A Corpus and Method for News Multi-Headline Generation Dayiheng Liu♠∗ Yeyun Gong† Yu Yan‡ Jie Fu♦ Bo Shao† Daxin Jiang‡ Jiancheng Lv♠ Nan Duan† ♠College of Computer Science, Sichuan University †Microsoft Research Asia ♦ Mila ‡Microsoft losinuris@gmail.com Abstract News headline generation aims to produce a short sentence to attract readers to read the news. One news article often contains mul- tiple keyphrases that are of interest to dif- ferent users, which can naturally have multi- ple reasonable headlines. However, most ex- isting methods focus on the single headline generation. In this paper, we propose gener- ating multiple headlines with keyphrases of user interests, whose main idea is to gener- ate multiple keyphrases of interest to users for the news first, and then generate multi- ple keyphrase-relevant headlines. We propose a multi-source Transformer decoder, which takes three sources as inputs: (a) keyphrase, (b) keyphrase-filtered article, and (c) original article to generate keyphrase-relevant, high- quality, and diverse headlines. Furthermore, we propose a simple and effective method to mine the keyphrases of interest in the news article and build a first large-scale keyphrase- aware news headline corpus, which contains over 180K aligned triples of ⟨news article, headline, keyphrase⟩. Extensive experimental comparisons on the real-world dataset show that the proposed method achieves state-of-the- art results in terms of quality and diversity1. 1 Introduction News Headline Generation is an under-explored subtask of text summarization (See et al., 2017; Gehrmann et al., 2018; Zhong et al., 2019). Un- like text summaries that contain multiple context- related sentences to cover the main ideas of a document, news headlines often contain a single short sentence to encourage users to read the news. Since one news article typically contains multiple keyphrases or topics of interest to different users, Asia.∗ Work is done during internship at Microsoft Research 1The source code will be available at https://github.com/ dayihengliu/KeyMultiHeadline. it is useful to generate multiple headlines covering different keyphrases for the news article. Multi- headline generation aims to generate multiple inde- pendent headlines, which allows us to recommend news with different news headlines based on the in- terests of users. Besides, multi-headline generation can provide multiple hints for human news editors to assist them in writing news headlines. However, most existing methods (Takase et al., 2016; Ayana et al., 2016; Murao et al., 2019; Col- menares et al., 2019; Zhang et al., 2018) focus on single-headline generation. The headline genera- tion process is treated as an one-to-one mapping (the input is an article and the output is a head- line), which trains and tests the models without any additional guiding information or constraints. We argue that this may lead to two problems. Firstly, since it is reasonable to generate multiple head- lines for the news, training to generate the single ground-truth might result in a lack of more detailed guidance. Even worse, a single ground-truth …

Content truncated to 500 words