Edinburgh Research Explorer

Whodunnit? Crime Drama as a Case for Natural Language Understanding

Research output: Contribution to journalArticle

Related Edinburgh Organisations

Open Access permissions

Open

Documents

  • Download as Adobe PDF

    Accepted author manuscript, 1 MB, PDF document

    Licence: Creative Commons: Attribution (CC-BY)

  • Download as Adobe PDF

    Final published version, 979 KB, PDF document

    Licence: Creative Commons: Attribution (CC-BY)

https://www.transacl.org/ojs/index.php/tacl/article/view/1282
Original languageEnglish
Pages (from-to)1-15
Number of pages15
JournalTransactions of the Association for Computational Linguistics
Volume6
DOIs
Publication statusPublished - 11 Jan 2018

Abstract

In this paper we argue that crime drama exemplified in television programs such as CSI: Crime Scene Investigation is an ideal testbed for approximating real-world natural language understanding and the complex inferences associated with it. We propose to treat crime drama as a new inference task, capitalizing on the fact that each episode poses the same basic question (i.e., who committed the crime) and naturally provides the answer when the perpetrator is revealed. We develop a new dataset1 based on CSI episodes, formalize perpetrator identification as a sequence labeling problem, and develop an LSTM-based model which learns from multi-modal data. Experimental results show that an incremental inference strategy is key to making accurate guesses as well as learning from representations fusing textual, visual, and acoustic input.

Download statistics

No data available

ID: 50778397