Computational narrative is a complex field. While the computational processing of narratives has been tackled from diferent perspectives, the literature has focused on the analysis of a main plot, even if it is composed of diferent narrative sub-plots. However, modern narrative often involves additional threads that complement, justify or enhance the main narrative. This is commonly observed in narrative in video games. In this paper, we present a computational engine that can automatically diferentiate between primary and secondary stories. This diferentiation is based on the identification of a number of features that are believed to be key factors in the classification. We have analyzed a number of stories by using a simple game prototype with several stories, and a prototype assessment of the efectiveness of the system with human players has been carried out. Results indicate the hypothesized features can certainly be used to identify or produce secondary stories linked to a main plot.
There exist many eforts towards achieving rich narrative generation, particularly in
videogames[
In order to diferentiate the narratives that make up a video game into main and secondary
ones,this research proposes to represent stories and narratives in a classic way, observing the
writing style of certain novel authors who structure their stories through chapters that are in
turn composed of diferent scenes using the following structure: Story - Chapters - Scenes [
In order to test the hypothesis, an simple videogame in which to develop the plot has been developed. This environment includes a main story together with the necessary requirements, and several secondary stories. The content of the stories was given the same amount of semantic relevance in the story. Only the values of chapters, characters and places were diferent. The story was implemented in a simple prototype, using a 2D style with RPG modality to develop each of the plots in a setting set in medieval fantasy. Only two small cities, a forest and a dungeon that connects them and a maze were present in the prototype. In addition, the game implements some systems necessary to progress, such as the player’s interaction with the game elements and control system, as well as managing the progress of the stories by starting diferent dialogues, advancing between the diferent chapters of the plots. The main story is loaded and executed from the beginning, while the secondary ones are executed by interacting with specific elements of the setting, such as some NPCs or some object. In order for the results of the experiment to be as objective as possible, the game was organized in a way that the players would not know which was the main story and which ones were the secondaries. Therefore, all stories started at the same town talking to diferent NPCs and triggering their respective storyline. A total of 10 people (3 women and 7 men, from 19 to 63 years old) participated in the experiment. The participants were given access to a questionnaire gathering the time spent in the game; perception of the game as a whole, such as the number of stories he/she has been able to identify and which of all those he/she considers as main; and analysis of each of the game to know how involved the variables of the hypothesis are in identifying stories. For the ifrst story, which was designed to be the main narrative for this prototype, 9 out of 10 players found it and completed it and 8 of them considered it as the main narrative. Then they were asked for more information and why they considered this story as the main narrative of the game. The reasons why this was considered were because of the narrative importance, the characters involved in it and the objects that took part on it. It is also relevant to point out that 7 out of the 8 people who considered it the main narrative, was because this story is the first one to be played in the game. As for the length of the story it was considered to be of a medium duration. Most of the people found it to have a normal length. As for the number of characters involved it was also considered to be of a normal quantity. The second story was played by 9 out of the 10 total players. However, none of the players that found and played this story considered it to be the main narrative of the game. When asked if this story could be considered as the main narrative only 22.2% of the players considered it to be relevant enough to work as a main narrative plot. The people who considered it said that it was the most developed story, while the rest of the players said that it was too short to be the main plot. It was mentioned as well that this story’s ending is not very open and is hard for it to be continued. For the duration of the story, as mentioned before, it was considered by most of the players to be too short. As for the characters quantity, it was considered to be enough and normal. In he third story only 70% of players interacted with it, making this story to be not very well situated for the players to locate it. Also, it was not considered to be the main narrative by all of the testers who found it, making it’s design as a secondary narrative correct. However, contrary to the second one, 100% of the players said that it couldn’t be the main plot of the game. One of the arguments given for this was that there was too many characters and with little narrative relevance, which gives the result that it does not matter the quantity of the characters, but the narrative importance of those characters. This will be important because it will afect the ifnal equation. The fourth story, it was played by the 80% of the players, which made it the second most played secondary story. As for the classification of the story, seven out of the eight players considered it as a secondary narrative. However, three out of eight said that it could be considered as the main narrative. This was because of the open ending and the possibility to continue it. As for the length of the story and the quantity of characters it was considered to be of a medium-short length with a small quantity of characters. This reinforced the idea that the quantity of characters was not that relevant against the narrative relevance. Table 1 shows the results.
Chapter 33 33 22.2 11.1
Characters Places 26.6 50 9.39 42.85 37.3 28.5 22.8 28.5
An automatic story categorization model, capable of diferentiating between main and secondary stories has been developed, and a simple experiment has been run. The results are partially positive and the model is scalable, but the experiments are limited by the number of participants and the impact of the hand-written narratives. Other aspects could also help to refine the approach. For instance as the importance of objects, which according to the surveys carried out with the players in both phases of the experiment could be just as important as the variable centered on the importance of characters. It was observed that stories with open endings have also captured the attention of respondents, who considered that the stories could continue and be much larger than what was seen in the prototype. Finally, the location of the character that starts the story is also very relevant, since if the NPC in charge of said task is near the player once the game starts, the player has to relate it to the main story.