Faculty Sponsor: Yuxuan Mei
Dylan Ross
My name is Dylan Ross, a rising junior (’28) at Wesleyan University studying Mathematics, Music, and Computer Science. Once I graduate, I hope to have enough time outside of work to focus on my personal projects: composing albums, programming games, and drawing whatever comes to mind.
Abstract: Ravelry is the largest knitting website, composed of tens of thousands of community contributed creations in several languages. Over the past few months, my research project’s goal was to create a predictive model using the website’s vast selection of instructions. Specifically, my model can predict the construction labels that would be associated with any set of notes, including pdfs. This was achieved after vigorous pre-processing and using TF-IDF vectorizers and logistic regression. The model can also be easily trained to predict other attributes, including but not limited to pattern types (i.e. ‘is_toy’ vs ‘is_garment’). The next logical steps of this project are to compare this model to zero-shot classification by an LLM and to implement this model as a feature in Ravelry, allowing users to automatically select tags when publishing their works.
text_mining_poster