Adaptive interactive storytelling is a resource used to guide the player through the game narrative in a way he/she can understand, in his/her own unique way, the multifaceted nature of the story being told. In interactive storytelling, the player is presented with a set of options to help directing the narrative, and the Drama Manager (DM) is responsible for controlling the narrative trajectories the user will experience. This work proposes Aldabra, a DM based on a genetic algorithm that is able to present non-deterministic plot lines to the player according to his/her model. The proposed DM will follow the Anthropological Structures of the Imaginary proposed by Durand to guide the narrative, which will be the basis for a player model. Experiments were performed both with synthetic datasets as well as with a group of players, evaluating important properties such as Adaptability, Replayability, Authorial Control and Player Autonomy. The quantitative results show that the GA finds subsets of plot lines that are not far from the optimum while increasing replayability. Qualitative analyses of the players preference considering the three regiments that defined Durand’s theory demonstrate that the theory is appropriate to model the players preferences in interactive storytelling.

自适应交互式故事讲述是一种资源，用于引导玩家以自己的独特方式理解故事的多面性，从而通过其游戏故事来进行指导。在交互式故事讲述中，向播放器提供了一组选项来帮助指导叙事，而Drama Manager（DM）负责控制用户将经历的叙事轨迹。这项工作提出了Aldabra，这是一种基于遗传算法的DM，能够根据玩家的模型向其展示不确定的情节线。拟议的发展议程将遵循虚构的人类学结构由杜兰德（Durand）提出以指导叙述，这将成为玩家模型的基础。对合成数据集和一组参与者都进行了实验，评估了重要的属性，例如适应性，可重放性，授权控制和玩家自治。定量结果表明，GA在提高可重现性的同时，发现了与最优线相距不远的情节线子集。考虑定义杜兰德理论的三个方案的玩家偏好的定性分析表明，该理论适合在交互式故事讲述中为玩家偏好建模。