News Article Teaser Tweets and How to Generate Them

21894

Authors

nan

Abstract

AbstractIn this work, we define the task of teaser generation and provide an evaluation benchmark and baseline systems for the process of generating teasers. A teaser is a short reading suggestion for an article that is illustrative and includes curiosity-arousing elements to entice potential readers to read particular news items. Teasers are one of the main vehicles for transmitting news to social media users. We compile a novel dataset of teasers by systematically accumulating tweets and selecting those that conform to the teaser definition. We have compared a number of neural abstractive architectures on the task of teaser generation and the overall best performing system is See et al. seq2seq with pointer network.

Keywords

nan

Full Content

Proceedings of NAACL-HLT 2019, pages 3967–3977 Minneapolis, Minnesota, June 2 - June 7, 2019. c⃝2019 Association for Computational Linguistics 3967 News Article Teaser Tweets and How to Generate Them Sanjeev Kumar Karn1,2, Mark Buckley2, Ulli Waltinger2 and Hinrich Sch¨utze1 1Center for Information and Language Processing (CIS), LMU Munich 2Machine Intelligence, Siemens CT, Munich, Germany 1skarn@cis.lmu.de 2{sanjeev.kumar karn,mark.buckley,ulli.waltinger}@siemens.com Abstract In this work, we define the task of teaser gen- eration and provide an evaluation benchmark and baseline systems for the process of gen- erating teasers. A teaser is a short reading suggestion for an article that is illustrative and includes curiosity-arousing elements to entice potential readers to read particular news items. Teasers are one of the main vehicles for trans- mitting news to social media users. We com- pile a novel dataset of teasers by systemati- cally accumulating tweets and selecting those that conform to the teaser definition. We have compared a number of neural abstractive ar- chitectures on the task of teaser generation and the overall best performing system is See et al. (2017)’s seq2seq with pointer network. 1 Introduction A considerable number of people get their news in some digital format.1 The trend has made many publishers and editors shift their focus to the web and experiment with new techniques to lure an Internet-savvy generation of readers to read their news stories. Therefore, there has been a notice- able increase in the sharing of short illustrative pieces of texts about the news on social media. We define a ShortcutText as a short text (about 15 words or less) describing and pointing to a news article and whose purpose is to invite the recipi- ent to read the article. A headline is a Shortcut- Text that optimizes the relevance of the story to its reader by including interesting and high news value content from the article (Dor, 2003). Click- bait is a pejorative term for web content whose main goal is to make a user click an adjoining link by exploiting the information gap. According to the definition, a principal part of the headline is an 1http://www.journalism.org/2008/07/21/the-influence-of- the-web/ extract of the article, thereby creating an impres- sion of the upcoming story. However, click-bait, a ShortcutText, contains mostly elements that cre- ate anticipation, thereby making a reader click on the link; however, the reader comes to regret their decision when the story does not match the click- bait’s impression (Blom and Hansen, 2015). Thus, click-bait provides a false impression (non-bona fide) and contains insufficient information (highly abstractive). bona-fide teasing abstractive headline yes no low clickbait no yes high teaser yes yes high Table 1: The table shows three categories of Shortcut- Texts and their properties We introduce the new concept of teaser and define it as a ShortcutText devised by fusing curiosity-arousing elements with interesting facts from the article in a manner that concurrently cre- ates a valid impression of an upcoming story and a sense of incompleteness, which motivates the au- dience to read the article. A teaser is one …

Content truncated to 500 words

Labeled Text

AbstractIn this work, we define the task of teaser generation and provide an evaluation benchmark and baseline systems for the process of generating teasers. A teaser is a short reading suggestion for an article that is illustrative and includes curiosity-arousing elements to entice potential readers to read particular news items. Teasers are one of the main vehicles for transmitting news to social media users. We compile a novel dataset of teasers by systematically accumulating tweets and selecting those that conform to the teaser definition. We have compared a number of neural abstractive architectures on the task of teaser generation and the overall best performing system is See et al. seq2seq with pointer network.

Publication Details

Paper ID:
21894
Year:
2019
Publication:
TIPSTER
Publication Type:
conf
DOI:
nan