Registration for the COLING2022 #CreativeSum
Workshop Shared Tasks (books & tv/movie script summarization)
is now open, the form is available here: https://bit.ly/3LFBXC4
Registration closes July 1st.
https://creativesumm.github.io/
Automatic Summarization for Creative Writing
Hybrid-Mode Workshop
COLING 2022 at Gyeongju, Republic of Korea, October, 2022
Overview
Text summarization aims at condensing long documents into short paragraphs that include salient information. Given the constantly growing volume of online documents, automatic text summarization can help people to find information relevant to their interests. We envision that summarization systems of the future will need to be equipped with the ability to:
* process long input sequences spanning up to hundreds of pages of text
* analyze complex discourse structure such as narrative and multi-party dialog
* interpret figurative language to understand and convey the salient points in the input
Most research in the field has been done in the newswire and scientific domains. While important, these domains pose limited challenges for future generations of summarization due to the limited input length, literal and/or technical language, positional biases, and constrained discourse structure.
An equally important, yet underexplored, domain for text summarization is creative writing, which includes documents such as books, stories, as well as scripts from plays, TV shows, and movies. Documents in this domain are uniquely characterized by their substantial length, non-trivial temporal dependencies (e.g., parallel plot threads and non-linear plot development), complex structures which often combine narrative and multi-party dialogs, and a wide variety of styles. Successfully summarizing such texts requires making literary interpretations, conveying implicit information, and heavily paraphrasing the input. This makes summarizing creative documents a challenging task requiring techniques that have not yet been explored in the field.
Invited Speakers:
Mirella Lapata, Edinburgh
Asli Celikyilmaz, Facebook AI
Shashi Narayan, Google AI
Greg Durrett, UT Austin
Mohit Bansal, UNC
Miguel Ballesteros, Amazon
Lu Wang, Michigan
Xiaojin Wan, Peking University