MPST: A Corpus of Movie Plot Synopses with Tags

MPST: A Corpus of Movie Plot Synopses with Tags Sudipta Kar, Suraj Maharjan, A. Pastor L´opez-Monroy and Thamar Solorio Department of Computer Science University of Houston Houston, TX 77204-3010 {skar3, smaharjan2, alopezmonroy, tsolorio}@uh.edu Abstract Social tagging of movies reveals a wide range of heterogeneous information about movies, like the genre, plot structure, soundtracks, metadata, visual and emotional experiences. Such information can be valuable in building automatic systems to create tags for movies. Automatic tagging systems can help recommendation engines to improve the retrieval of similar movies as well as help viewers to know what to expect from a movie in advance. In this paper, we set out to the task of collecting a corpus of movie plot synopses and tags. We describe a methodology that enabled us to build a fine-grained set of around 70 tags exposing heterogeneous characteristics of movie plots and the multi-label associations of these tags with some 14K movie plot synopses. We investigate how these tags correlate with movies and the flow of emotions throughout different types of movies. Finally, we use this corpus to explore the feasibility of inferring tags from plot synopses. We expect the corpus will be useful in other tasks where analysis of narratives is relevant. Keywords: Tag generation for movies, Movie plot analysis, Multi-label dataset, Narrative texts 1. Introduction Folksonomy (Vander Wal, 2005), also known as collabora- tive tagging or social tagging, is a popular way to gather community feedback about online items in the form of tags. User-generated tags in recommendation systems like IMDb1 and MovieLens2 provide different types of summarized attributes of movies. These tags are effective search keywords, are also useful for discovering social interests, and improving recommendation performance (Lambiotte and Ausloos, 2006; Szomszor et al., 2007; Li et al., 2008; Borne, 2013). In this regard, an interesting research question is: Can we learn to predict tags for a movie from its written plot synopsis? This question enables an enormous potential to understand the properties of plot synopses that correlate with the tags. For instance, a movie can be tagged with fantasy, murder, and insanity, that represent different summarized attributes of the movie. The inference of multiple tags by analyzing the written plot synopsis of movies can benefit the recommendation engines. In addition, the consumers would have a useful set of tags representing the plot of a movie. Notwithstanding the usefulness of tags, its proper use in computational methods is challenging as the tag spaces are noisy and redundant (Katakis et al., 2008). Noise and redundancy issues arise because of differences in user perspectives and use of semantically similar tags. For example, the Movielens 20M dataset (Harper and Konstan, 2016), which provides tag assignments between ≈27K movies and ≈1,100 unique tags also suffers from these problems. Thus, a fine-grained tagset and their assignment to movie plots can help to overcome these obstacles. In this work, (i) we present the MPST corpus that contains plot synopses of 14,828 movies and their associations with a set of 71 fine-grained tags; …

Content truncated to 500 words