Semiotic Structuring in Movie Narrative Generation

In this paper we apply, in a novel way, our ongoing research work on the interactive composition of narratives based on semiotic relations. To the two basic components of interactive systems, namely, a software tool and a user interface, we add a third component - AI agents, understood as an upgraded rendition of software agents. Our semiotic relations approach considers four ways of composing new narratives from existing narratives. Along the horizontal syntagmatic axis one can form the new narrative by combining two or more previous narratives. Along the vertical paradigmatic axis, the new narrative may emerge as a similar version, which imitates the previous one in a different context. Along the depth meronymic axis, the hierarchic narrative levels, such as event and scene, are explored, allowing to zoom in and out in the composition process. Lastly, the antithetic consideration, rather than adding a dimension, aims at some form of reversal, through the adoption of opposite values. A fully operational prototype is described, with ChatGPT operating as the main AI agent component. To run the experiments, we concentrated on movie narratives.