Generador de historias basado en agentes
Loading...
Official URL
Full text at PDC
Publication date
2014
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citation
Abstract
En este proyecto discutiremos cómo podemos generar creatividad computacional. Esto es una tarea compleja, ya que el propio concepto de lo que es la creatividad es ambiguo, puesto que en él intervienen cosas como el talento, que son de por sí difíciles de explicar. Siguiendo el dominio de nuestra aplicación, nos concentraremos en el ámbito de la narrativa y la creación de historias. Veremos cómo podemos generar historias automáticamente, además de los diferentes generadores
con los que nos hemos encontrado, y que nos han servido como fuente de inspiración en mayor o
menor medida. Basándonos en la generación de historias por simulación y planificación, hemos implementado un generador de historias basado en agentes inteligentes para conseguirlo. Estos agentes utilizarán un planificador externo para saber qué es lo que tienen que hacer en cada momento, lo que dará lugar a interacciones entre los mismos. Utilizaremos estas interacciones entre agentes como contenido de las historias a generar, de forma que en cierto modo, serán estos agentes los que nos hagan parte del trabajo. Recogiendo los datos producidos por estos agentes, habremos terminado la fase de "crear", y ya sólo nos quedará la fase de "narrar". Para ello, por el momento, tendremos textos predefinidos dentro del propio código, aunque la idea inicial era utilizar un programa externo que tomase como entrada un log con las interacciones producidas. Con todo, esto, hemos logrado implementar un generador de historias básico, capaz de generar historias distintas e independientes unas de otras.
In this project we discuss about how we can generate computational creativity. This is a complex work, as the concept of creativity itself is ambiguous, inasmuch as there are things such as talent involved in it, which are difficult to explain themselves. Following the domain of our application, we concentrate in the scope of narrative and story generation. We’ll see how we can generate storiesautomatically, in addition to the different storytellers that we’ve come across, in which our project relies to a greater or lesser extent. Relying on story generation based on intelligent agents an planification, we’ve implemented our own storyteller. These agents use an external planner to know what to do whenever they want, which leads to interactions between them. We use this interactions between agents as the content of the stories generated, so that, in some way, these agents will make part of the work for us. Once we’ve collected the data produced by the agents, we’d have finished the “create” phase, and there will only be the “narrate” phase remaining. For this, by now, we’ve predefined text in between the code itself, although our initial idea was to use an external program, that had the log with the interactions occurred as an input. With all this, we’ve achieved to implement a basic storyteller, capable of generating not only different, but independent stories.
In this project we discuss about how we can generate computational creativity. This is a complex work, as the concept of creativity itself is ambiguous, inasmuch as there are things such as talent involved in it, which are difficult to explain themselves. Following the domain of our application, we concentrate in the scope of narrative and story generation. We’ll see how we can generate storiesautomatically, in addition to the different storytellers that we’ve come across, in which our project relies to a greater or lesser extent. Relying on story generation based on intelligent agents an planification, we’ve implemented our own storyteller. These agents use an external planner to know what to do whenever they want, which leads to interactions between them. We use this interactions between agents as the content of the stories generated, so that, in some way, these agents will make part of the work for us. Once we’ve collected the data produced by the agents, we’d have finished the “create” phase, and there will only be the “narrate” phase remaining. For this, by now, we’ve predefined text in between the code itself, although our initial idea was to use an external program, that had the log with the interactions occurred as an input. With all this, we’ve achieved to implement a basic storyteller, capable of generating not only different, but independent stories.
Description
Trabajo de Fin de Grado en Ingeniería Informática (Universidad Complutense, Facultad de Informática, curso 2013/2014)