Case Study
Authors: Angelo Ciarlini, Oliver Schneider, Michael Straeubig, Emma Whittaker, David Pizzi, Stefan Rank
This case study is the result of a workshop with the title "Modelling a storyworld based on planning" which was conducted during the Summerschool 2011 in Wiesbaden, Germany and was finally implemented in the EmoEmma system.
Story
The final story “Diver’s Luck” circles around Vladimir Putin’s discovery of two ancient Greek urns during a scuba-diving expedition shortly before the beginning of the “Summerschool” in mid august. At this time it was unclear if that was just a publicity hoax, which was indeed revealed to be the case a few weeks later. Not knowing that, the story’s content speculates about possible reasons for the discovery and how they could influence the upcoming elections in Russia. There are multiple possible outcomes, for example depending on whether the urns were actually placed or truly discovered and whether that event was witnessed by someone who could report on it.
Outcome
The figure below on the left side shows one outcome of the story in the final visualisation, in particular the shortest one possible. The text on the right is accompanied by a corresponding picture on the left. The showing of the text is triggered by the action which was previously chosen by the planning algorithm. This also shows that an action in the sense of planning and IDS can have a more abstract meaning and includes a lot of actions on a lower but, in the case of this story, only representational level (like “Alice Bugov gets called”). The figure on the right shows an alternative story which is the result of user interaction.
 |
 |
The picture below shows the final presentation of the workshop results.
 |
Creation Process
The pictures below show some basic assumptions and character constellations, demonstrating the common use of graphical representations of story elements, like scenario and character maps. The use of graphical representations, especially regarding the relations between characters and regarding their background stories was an important tool to connect the abstract and in principle domain-independent planning formalism to story ideas.
 |
 |
 |
The creation process started with the collection of different story ideas. After a short introduction to the EmoEmma authoring tool and the specific formalism for planning, the decision was made to use a newspaper article regarding Putin’s discovery as a base for the story. This decision itself was not directly motivated by the specifics of the available formalism, but it proved to be very useful to have a story-world that is anchored by real-world events with a clear causal structure. One of the most important clarifications during the creation process was the fact that the underlying planning formalism is completely domain-independent and also does not assume a specific form of user interaction. In practice, this meant that the group had to decide on their own at what level of granularity story events should be represented and how user interaction should, in principle, look like.
This independence turned out to be both a problem and a boon, respectively, for different styles of authoring that were represented in the group. For practical reasons, i.e. to restrict the time necessary for formalising the story domain, a rather high-level representation of actions and facts was chosen, and the type of user interaction was left open apart from indicating specific actions as involving user choice.
Already during these basic discussions, different possible actions (abstract story events) were tried out to see if they would suit the purpose of creating alternative orders of events. Initially, pen and paper and flipchart were used to prototype different action sequences. The picture below shows some considerations during that process also regarding the relation between “actions” and “facts” (or propositions), user interaction and goal states.
 |
In the next step, said facts had to be defined so that the described storyworld would allow all authored actions and intended action sequences as unambiguous paths, by setting suitable preconditions and effects for the actions. Small cards were used as an intermediate step for fleshing out the action sequences. However, at this stage, the tool itself already proved useful to visualise potential paths and especially for identifying missing facts to distinguish between different paths.
The pictures below are giving a general impression of the work process and show discussions circling around the definitions of these actions and facts.
 |
 |
 |
Authoring of the actual formalisation was performed mostly by the author of the tool itself, predominantly in order to save time, as he was very familiar with the details of correctly specifying actions, preconditions and effects. Further, the group decided to split the efforts, so that part of the group could focus on fleshing out the backstory of the characters as well as the textual and graphical representation of story events, while the rest focused on the correct formalsation of story plans. This division also resulted in a very fruitful and enlightening back-and-forth between the two efforts, mainly regarding story details that were not represented in the level of granularity chosen for the plan formalisation.
Downloads
Documents
The following material is available available in the Internal Download Area (registration and log-in needed)
- EmoEmma Authoring Tool Extended ReadMe (PDF)
- EmoEmma Authoring Tool Manual (PDF)
- EmoEmma Authoring Tool Presentation (PDF)
Software
The following material is available on request from David Pizzi: D.Pizzi(at)tees(dot)ac(dot)uk
- EmoEmma Authoring Tool incl. "Diver's Luck" (ZIP, incl. ReadMe.txt for quick setup)
