TOP
Suche auf der Schloss Dagstuhl Webseite
Sie suchen nach Informationen auf den Webseiten der einzelnen Seminare? - Dann:
Nicht fündig geworden? - Einige unserer Dienste laufen auf separaten Webseiten mit jeweils eigener Suche. Bitte beachten Sie folgende Liste:
Schloss Dagstuhl - LZI - Logo
Schloss Dagstuhl Services
Seminare
Innerhalb dieser Seite:
Externe Seiten:
  • DOOR (zum Registrieren eines Dagstuhl Aufenthaltes)
  • DOSA (zum Beantragen künftiger Dagstuhl Seminare oder Dagstuhl Perspektiven Workshops)
Publishing
Innerhalb dieser Seite:
Externe Seiten:
dblp
Innerhalb dieser Seite:
Externe Seiten:
  • die Informatik-Bibliographiedatenbank dblp


Dagstuhl-Seminar 19172

Computational Creativity Meets Digital Literary Studies

( 22. Apr – 25. Apr, 2019 )

(zum Vergrößern in der Bildmitte klicken)

Permalink
Bitte benutzen Sie folgende Kurz-Url zum Verlinken dieser Seite: https://www.dagstuhl.de/19172

Organisatoren

Kontakt



Programm

Motivation

Stories are complex webs connecting imaginary events and people, which are written and told to entertain. The corresponding art of conceiving the most entertaining plots in which these events and people play roles has not only been tackled by storytellers but has recently also been approached using computational means. Still, Computational Storytelling (CS) as a prominent subfield within Computational Creativity (CC) hitherto has mostly focused on planning stories and creating them with the help of text generation methods from Natural Language Processing (NLP), thus simulating a logically coherent plot. A good story, however, is much more than that. Relevant aspects also include narrative concepts like narrative style, chronology of narratives, focalization and perspective – which currently to most parts still lie beyond the horizon of CS. This does not hold for every discipline, though: These narratological concepts have been investigated by literary scholars for a long time.

Digital methods, in turn, have recently been of interest as an approach towards literature analysis, often referred to as Digital Literary Studies (DLS). Yet, operationalization of these well-understood concepts is required when used as the basis for computational modelling – a challenge that is often taken up in collaborations between researchers from NLP and the Literary Studies. The process of making concepts explicit enough to operationalize them can in turn sharpen the definitions of theoretical considerations and feed back into theoretical discussions.

These are just a few of the obvious meeting points between CC/CS and NLP and between NLP and DLS. However, these connections are not (yet) transitive. The goal of this Dagstuhl Seminar is to overcome the current divisions and establish links between all three disciplines and among involved researchers.

The potential benefits for the respective fields include:

  • One of the major challenges in DLS is the approximation of concepts with computational approaches (i.e. their operationalization) which not only requires a translation of the concepts, but also a deep understanding of the deployed computational approaches used. This gap can best be tackled by providing expertise from the fields concerned. Whereas NLP is already accepted as such a field (but still needed), CS has not been taken into consideration yet. Additionally, a second type of collaboration that still needs to be intensified is the one that connects the interpretative, manual annotations from DLS with computational approaches to text analysis and generation.
  • In its early days, NLP has focused on a limited variety of texts and consequently suffers from a bias towards newspaper texts. Even though there are efforts towards more diverse and flexible text processing, the constant lack of data is a hindering factor. Digital Humanities (DH) and CC offer varieties of texts (and the potential means to generate further examples) mitigating the impasse.
  • In CC, CS focuses almost exclusively on plot and logical structure of storytelling. However, as mentioned before, a narrative is a complex web of different factors that are well-investigated in classical disciplines. While much work is based on formalist theories about narrative, other approaches from narrative theory still need to be explored better. For example, CS could benefit from the well-established fields of semiotics and structuralism as well as from more recent, reader-oriented developments in cognitive and empirical narratology.

In short, establishing the suggested links promises significant advantages for each of the involved fields, as all three focus on the same object of investigation. Among others, each community can gain insights into how to improve their own methods and findings without necessarily fully unifying the respective research fields. Bringing together both thought leaders and young aspiring researchers from the relevant communities, the seminar creates the necessary environment to establish professional and personal ties, forming the foundation for the required collaborations across disciplinary boundaries.

Copyright Tarek Richard Besold, Pablo Gervás, Evelyn Gius, and Sarah Schulz

Summary

Literary studies (LS) is a subfield of the humanities that provides a diversity of possible views on its objects of investigation. The universal approach to literary texts does not exist, instead there are many, sometimes incompatible theories that can be applied for the interpretation of literary texts. Additionally, with the emerging of the Digital Humanities (DH) the deployment of computational methods has been introduced into LS, leading to a further expansion of the range of theories and methodologies of text analysis and interpretation. Against that backdrop in the last decade much effort has especially been put into developing approaches that cover rather complex concepts for text analysis, including, among other, network theory (e.g., [7]) and approaches from distributional semantics for topic modelling and word vector estimation (e.g., [8]). This considerably changed the prerequisites of DH research in the field of LS. In many cases it is no longer possible to simply apply a predefined tool or algorithm, requiring traditionally trained LS scholars to move away from their disciplinary paradigm of individual research and towards adapting collaborative modes that can provide both LS and computational expertise. Researchers in Natural language processing (NLP) have shown considerable interest in text-based DH research. This interest is not only motivated by the diversity and complexity of the research questions, which offers an ideal testbed for the development of new methods and combined workflows, but also by the nature of texts found in the context of these research questions which are often diverse with respect to their lexical and syntactic range – meeting the need for this type of data in work aiming for more flexible NLP approaches. Computational Creativity (CC) is a multidisciplinary endeavour, modelling, simulating or replicating aspects of creativity using a computer, in order to achieve one of several ends: Either to construct a program or computer capable of human-level creativity, or to better understand human creativity and to formulate an algorithmic perspective on creative behaviour in humans, or to design programs that can enhance human creativity without necessarily being creative themselves (a concise overview of the main aspects of the field has, for instance, been laid out by [1]). One of CC’s most popular subfields is Computational Storytelling (CS), where researchers hitherto have mainly thought about the structure and logical implications of building blocks of stories, leaving most other dimensions of narrative construction out of consideration.

Taking stock of this overall state of affairs and the specific situation in the respective fields, the seminar was constructed around several main challenges:

  • One of the major challenges in DLS is the approximation of concepts with computational approaches to, i.e. their operationalization, that not only requires a translation of the concepts, but also a deep understanding of the deployed computational approaches used. This gap can be tackled best by providing expertise from the fields concerned. Whereas NLP is already accepted as such a field (but still needed), CS has not been taken much into consideration yet. A second type of collaboration that still needs to be intensified is the one that connects the interpretative, manual annotations from DLS (e.g., [4]) with computational approaches to text analysis and generation.
  • NLP has focused on a limited variety of texts in its beginnings and suffers from a bias towards newspaper texts. Even though there are efforts towards more diverse and flexible text processing, the constant lack of data is a problem. DH and CC offer a variety of texts to improve this situation – but are hitherto underused in that capacity.
  • CC, CS focuses almost exclusively on plot and logical structure of storytelling. However, a narrative is a complex web of different factors that are well-investigated in classical disciplines. While much work is based on formalist theories about narrative (especially [9]), other approaches from narrative theory still need to be explored better. For example, CS could benefit from the well-established fields of semiotics (e.g., [5]) and structuralism (e.g., [3]) as well as from more recent, reader-oriented developments in cognitive and empirical narratology (e.g. [6]; [2]).

In order to make researchers from the participating communities a) aware of the challenges and the corresponding opportunities an interdisciplinary meeting like the seminar offered, and b) make them take advantage of these opportunities still on-site, the seminar was split between presentations from researchers describing their recent work and questions they wanted to highlight for the audience, and ``hackathon'' phases in which decidedly interdisciplinary teams of participants worked on concrete projects.

The following pages summarize the content of these presentations and the outcomes of the group projects.

References

  1. M. A. Boden. How computational creativity began. In M. Besold, T. R.; Schorlemmer and A. Smaill, editors, Computational Creativity Research: Towards Creative Machines. Atlantis Press, Amsterdam, 2015.
  2. M. Bortolussi, P. Dixon, and F. C. E.P. Dixon. Psychonarratology: Foundations for the Empirical Study of Literary Response. Psychonarratology: Foundations for the Empirical Study of Literary Response. Cambridge University Press, 2003.
  3. G. Genette. Narrative discourse:. G – Reference, Information and Interdisciplinary Subjects Series. Cornell University Press, 1980.
  4. E. Gius. In Diegesis, page 4. 2016.
  5. A. J. Greimas. Structural Semantics: An Attempt at a Method. University of Nebraska Press, 1983.
  6. D. Herman. Story Logic: Problems and Possibilities of Narrative. Frontiers of narrative. University of Nebraska Press, 2002.
  7. F. Moretti. Network Theory, Plot Analysis. Literary lab. Stanford Literary Lab, 2011.
  8. Christof Schöch. Topic Modeling Genre: An Exploration of French Classical nd Enlightenment Drama. Digital Humanities Quarterly, November 2016. This is a pre-publication version of an article to appear in Digital Humanities Quarterly. Last revision: October 2016.
  9. Vladimir Propp. Morphology of the Folktale.University of Texas Press, 2010.
19172
Copyright Sarah Schulz, Tarek Richard Besold, Pablo Gervás, and Evelyn Gius

Teilnehmer
  • Leonid Berov (Universität Osnabrück, DE) [dblp]
  • Tarek Richard Besold (Telefonica Innovacion Alpha - Barcelona, ES) [dblp]
  • Amilcar Cardoso (University of Coimbra, PT) [dblp]
  • João Miguel Cunha (University of Coimbra, PT) [dblp]
  • Thierry Declerck (DFKI - Saarbrücken, DE) [dblp]
  • Mark Finlayson (Florida International University - Miami, US) [dblp]
  • Pablo Gervás (Complutense University of Madrid, ES) [dblp]
  • Evelyn Gius (TU Darmstadt, DE) [dblp]
  • Marina Grishakova (University of Tartu, EE)
  • Christian Guckelsberger (Queen Mary University of London, GB) [dblp]
  • Christopher Hench (Amazon - Cambridge, US) [dblp]
  • Jonas Kuhn (Universität Stuttgart, DE) [dblp]
  • Kai-Uwe Kühnberger (Universität Osnabrück, DE) [dblp]
  • Oliver Kutz (Free University of Bozen-Bolzano, IT) [dblp]
  • Carlos León (Complutense University of Madrid, ES) [dblp]
  • Rafael Pérez y Pérez (Universidad Autonoma Metropolitana - Cuajimalpa, MX) [dblp]
  • Enric Plaza (CSIC - Bellaterra, ES) [dblp]
  • Manuel Portela (Universidade de Coimbra, PT) [dblp]
  • Dino Pozder (University of Tartu, EE)
  • Nils Reiter (Universität Stuttgart, DE) [dblp]
  • Sarah Schulz (Ada Health - Berlin, DE) [dblp]
  • Hannu Toivonen (University of Helsinki, FI) [dblp]
  • Sara L. Uckelman (Durham University, GB) [dblp]
  • Tony Veale (University College Dublin, IE) [dblp]
  • Philipp Wicke (University College Dublin, IE) [dblp]
  • Sina Zarrieß (Universität Bielefeld, DE) [dblp]

Klassifikation
  • artificial intelligence / robotics
  • multimedia
  • society / human-computer interaction

Schlagworte
  • computational creativity
  • storytelling
  • digital humanities
  • digital literary studies
  • computational narrativity