Document Type
Poster
Publication Date
Fall 2024
Abstract
The project starts by examining whether a smaller window size of 5% allows the algorithm to determine cruxes and their sentiment more accurately than the standard 10% window size, using the passages identified by the crux report to assist me. I find that the smaller window size seems to be more accurate for the most part, but this is not always the case. Moreover, regardless of window size, the algorithm struggled with definitively assessing ambivalent sentiment, much like how a reader has their own interpretation of emotionally charged moments. Moreover, using the results of the 5% and 10% window size comparisons, I discovered that the overall sentiment arc trajectory of the seven novels contradicts my own beliefs about the emotional arc of the novels. While as a reader, I find that the novels become progressively darker as Harry Potter matures and the conflict heightens, the algorithm believes otherwise, evaluating the first book as having an almost entirely negative sentiment while the latter books–particularly books 4-6–have an overall positive sentiment.
Recommended Citation
Newman, Xander and Elkins, Katherine, "Sentiment Analysis of Harry Potter: Exploring AI versus Human Interpretation" (2024). IPHS 200: Programming Humanity. Paper 72.
https://digital.kenyon.edu/dh_iphs_prog/72
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.