Please note: This PhD seminar will take place online.
Gaurav Sahu, PhD candidate
David R. Cheriton School of Computer Science
Supervisor: Professor Olga Vechtomova
Artistic inspiration is one of the least understood aspects of human creativity, yet it is crucial for the creation of works that resonate deeply with audiences. The complexity and unpredictability of stimuli and the vividness of poetic imagery are often key to evoking this inspiration.
In this work, we explore the computational modeling of artistic inspiration by identifying and analyzing key linguistic and poetic properties that contribute to what artists find inspiring in poetic and lyric lines. We introduce a dataset of annotated lyric lines, categorized as either “inspiring” or “not inspiring,” to facilitate the evaluation of these features. Our computational model leverages these properties to predict the inspirational quality of lyric lines. Our experiments show that our model achieves superior performance compared to a 450-shot classifier using LLaMA-3-70b, a state-of-the-art open-source language model. This work not only contributes a new dataset and feature set for lyric analysis but also advances the field by providing a more accurate tool for predicting artistic inspiration.