What’s This Movie About? A Joint Neural Network Architecture for Movie Content Analysis

Proceedings of NAACL-HLT 2018, pages 1770–1781 New Orleans, Louisiana, June 1 - 6, 2018. c⃝2018 Association for Computational Linguistics What’s this Movie about? A Joint Neural Network Architecture for Movie Content Analysis Philip John Gorinski and Mirella Lapata Institute for Language, Cognition and Computation School of Informatics, University of Edinburgh 10 Crichton Street, Edinburgh, EH8 9AB P.J.Gorinski@sms.ed.ac.uk, mlap@inf.ed.ac.uk Abstract This work takes a first step toward movie content analysis by tackling the novel task of movie overview generation. Overviews are natural language texts that give a first impres- sion of a movie, describing aspects such as its genre, plot, mood, or artistic style. We create a dataset that consists of movie scripts, attribute- value pairs for the movies’ aspects, as well as overviews, which we extract from an on- line database. We present a novel end-to-end model for overview generation, consisting of a multi-label encoder for identifying screen- play attributes, and an LSTM decoder to gen- erate natural language sentences conditioned on the identified attributes. Automatic and hu- man evaluation show that the encoder is able to reliably assign good labels for the movie’s attributes, and the overviews provide descrip- tions of the movie’s content which are infor- mative and faithful. 1 Introduction Movie summarization is the task of automatically summarizing a screenplay in order to gain a gen- eral impression of its content. This may include describing the movie’s main characters and plot, its genre, artistic style, and so on. As more and more movies are being produced every year1, there is an ever growing need to facilitate this task. Po- tential applications include producing shorter ver- sions of scripts to help with the decision mak- ing process in a production company, enhancing movie search by generating descriptions of what the movie is about, and notably, supporting movie recommendation engines by abstracting over spe- cific keywords to more general concepts. Figure 1 gives an example of the type of movie content analysis we would like to obtain auto- matically. The information is taken from Jinni, a 1According to http://www.boxofficemojo.com/ dur- ing 2009–2016, movie releases went up from 536 to 729. Mood: Suspenseful, Captivating, Tense, Scary Plot: Serial Killer, Special Agents, Investigation, Mind Game, Psychopath, Crimes, Deadly, Law Enforcement, Mind and Soul, Rivalry Genre: Crime, Thriller Style: Strong Female Presence Attitude: Serious, Realistic Place: Maryland, USA, Virginia Period: 20th Century, 90s Based on: Based on Book Praise: Award Winner, Blockbuster, Critically Ac- claimed, Oscar Winner, Modern Classic, Prestigious Awards Flag: Brief Nudity, Sexual Content, Strong Violent Content The Silence of the Lambs can be described as tense, captivat- ing, and suspenseful. The plot revolves around special agents, mind games, and a psychopath. The main genres are thriller and crime. In terms of style, The Silence of the Lambs stars a strong female character. In approach, it is serious and realis- tic. It is located in Maryland and Virginia. The Silence of the Lambs takes place in the 1990s. It is based on a book. The movie has received attention for …

Content truncated to 500 words