Elamur Research⌕
☰
home / papers / 2013 / Learning Latent Personas of Film Characters

Learning Latent Personas of Film Characters

Summary

The authors propose viewing narrative through its characters rather than only as a sequence of events. Their goal is to automatically discover the typical roles (personas) of film characters, that is, archetypes and stereotypes such as the villain or the protagonist's love interest. To this end, they assembled a new corpus of 42,306 movie plot summaries from the English-language Wikipedia and enriched it with metadata from Freebase: genres and the gender and age of the actors. The texts were processed with Stanford CoreNLP. For each character, they extracted a set of words of three types: actions the character performs, actions performed on the character, and attributes used to describe the character. Two generative latent-variable models are proposed. The Dirichlet Persona Model works with the text alone. The persona regression model additionally takes into account movie and actor metadata. Both models simultaneously group words into topics, topics into personas, and characters into personas. Inference is performed using collapsed Gibbs sampling. The authors regard this work as the first attempt to explicitly learn character personas. They openly release the data, the code, and a test set for evaluation, and the approach itself can be applied to literary fiction or to the analysis of media framing.

Key points

  • Corpus: 42,306 movie plots from a Wikipedia dump (November 2012). The median summary length is about 176 words, and about 715 for the 1,000 highest-grossing films. The metadata were taken from a Freebase dump, including 365 genre categories, which may overlap.
  • Features: each character is represented as a bag of (r, w) pairs. Here r is the role of the word (agent verb, patient verb or attribute), determined from Stanford CC-processed dependencies, and w is the lemma.
  • Filtering: the sample was restricted to the 1,000 most frequent events and to characters with at least 3 events. This yielded 120,345 characters. Of these, 33,559 could be linked to an actor of known gender, and 29,802 to an actor of known date of birth.
  • Actor demographics: among actors of known gender, about two thirds (66.7%) are men. The mean age at the time of the film's release is 37.9 years. Actresses are on average younger than actors: 33.0 versus 40.5 years.
  • Result: the model learned 100 latent personas. For example, the "action hero" cluster groups together Jason Bourne, Ethan Hunt and Peter Parker. The data and software are published at ark.cs.cmu.edu/personas.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages

Cited by

101 papers of our library cite it (of the 140673 whose bibliographies we read); 6 of them have a page here, 95 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work.

@inproceedings{bamman2013learning,
  title = {{Learning Latent Personas of Film Characters}},
  author = {Bamman, David and O'Connor, Brendan and Smith, Noah A.},
  booktitle = {Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2013},
  url = {https://aclanthology.org/P13-1035/}
}

References

29 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Aristotle. 335 BCE. Poetics, translated by Samuel H. Butcher (1902). Macmillan, London. find
  2. David M. Blei, Andrew Ng, and Michael Jordan. 2003. Latent dirichlet allocation. JMLR, 3:993–1022. find our card
  3. Joseph Campbell. 1949. The Hero with a Thousand Faces. Pantheon Books. find
  4. Nathanael Chambers and Dan Jurafsky. 2008. Unsupervised learning of narrative event chains. In Proceedings of ACL-08: HLT. find
  5. Nathanael Chambers and Dan Jurafsky. 2009. Unsupervised learning of narrative schemas and their participants. In Proceedings of the 47th Annual Meeting of the ACL. find
  6. Nathanael Chambers. 2011. Inducing Event Schemas and their Participants from Unlabeled Text. Ph.D. thesis, Stanford University. find
  7. Marie-Catherine de Marneffe and Christopher D. Manning. 2008. Stanford typed dependencies manual. Technical report, Stanford University. find
  8. Lajos Egri. 1946. The Art of Dramatic Writing. Simon and Schuster, New York. find
  9. Khalid El-Arini, Ulrich Paquet, Ralf Herbrich, Jurgen Van Gael, and Blaise Agüera y Arcas. 2012. Transparent user models for personalization. In Proceedings of the 18th ACM SIGKDD. find
  10. Micha Elsner. 2012. Character-based kernels for novelistic plot structure. In Proceedings of the 13th Conference of the EACL. find
  11. Mark Alan Finlayson. 2011. Learning Narrative Structure from Annotated Folktales. Ph.D. thesis, MIT. find
  12. R. A. Fisher. 1935. The Design of Experiments. Oliver and Boyde, Edinburgh and London. find
  13. Erving Goffman. 1959. The Presentation of the Self in Everyday Life. Anchor. find
  14. Amit Goyal, Ellen Riloff, and Hal Daumé, III. 2010. Automatically producing plot unit representations for narrative text. In Proceedings of the 2010 Conference on EMNLP. find
  15. Trond Grenager and Christopher D. Manning. 2006. Unsupervised discovery of a statistical verb lexicon. In Proceedings of the 2006 Conference on EMNLP. find
  16. Thomas L. Griffiths and Mark Steyvers. 2004. Finding scientific topics. PNAS, 101(suppl. 1):5228–5235. find
  17. Carl Jung. 1981. The Archetypes and The Collective Unconscious, volume 9 of Collected Works. Bollin-gen, Princeton, NJ, 2nd edition. find
  18. Neil McIntyre and Mirella Lapata. 2010. Plot induction and evolutionary search for story generation. In Proceedings of the 48th Annual Meeting of the ACL. Association for Computational Linguistics. find
  19. Robert McKee. 1997. Story: Substance, Structure, Style and the Principles of Screenwriting. Harper-Colllins. find
  20. Marina Meilă. 2007. Comparing clusterings—an information based distance. Journal of Multivariate Analysis, 98(5):873–895. find
  21. David Mimno and Andrew McCallum. 2008. Topic models conditioned on arbitrary features with dirichlet-multinomial regression. In Proceedings of UAI. find
  22. Fernando Pereira, Naftali Tishby, and Lillian Lee. 1993. Distributional clustering of english words. In Proceedings of the 31st Annual Meeting of the ACL. find
  23. E. J. G. Pitman. 1937. Significance tests which may be applied to samples from any population. Supplement to the Journal of the Royal Statistical Society, 4(1):119–130. find
  24. Michaela Regneri, Alexander Koller, and Manfred Pinkal. 2010. Learning script knowledge with web experiments. In Proceedings of the 48th Annual Meeting of the ACL. find
  25. Michaela Regneri, Alexander Koller, Josef Ruppenhofer, and Manfred Pinkal. 2011. Learning script participants from unlabeled data. In Proceedings of the Conference on Recent Advances in Natural Language Processing. find
  26. Roger C. Schank and Robert P. Abelson. 1977. Scripts, plans, goals, and understanding: An inquiry into human knowledge structures. Lawrence Erlbaum, Hillsdale, NJ. find
  27. Ivan Titov and Alexandre Klementiev. 2012. A bayesian approach to unsupervised semantic role induction. In Proceedings of the 13th Conference of EACL. find
  28. Greg C. G. Wei and Martin A. Tanner. 1990. A Monte Carlo implementation of the EM algorithm and the poor man’s data augmentation algorithms. Journal of the American Statistical Association, 85:699–704. find
  29. Limin Yao, Aria Haghighi, Sebastian Riedel, and Andrew McCallum. 2011. Structured relation discovery using generative models. In Proceedings of the Conference on EMNLP. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/papers/bamman-2013-learning-latent-personas-film-characters/