Metrics for desired structural features for narrative renderings of game logs

Abstract User performance in games is rarely recorded in formats amenable for easy revision. Game logs could be exploited as a source for reflection on user performance, with a view to improvement. Yet in their raw form they are difficult to interpret, and sometimes only specific parts of the game are worth reviewing. The ability to produce textual narratives that rework these logs (or the interesting parts of them) as stories could open up this wealth of data for further use. Existing work on narrative composition could be instrumental in achieving this end. This paper presents a model of the task of narrative composition as a set of operations that need to be carried out to obtain a linear sequence of event descriptions from a set of events that inspire the narration. An initial implementation of the model is applied to a chess game understood as a formalised set of events susceptible of story-like interpretations. Operating on simple representations of a chess game in algebraic notation, a number of candidate metrics are explored in relation to their variability over a number of games and their impact on the resulting tellings of a game.