Archive research has changed dramatically in form as the world has become more digital. Rather than poring through dusty stacks of books and papers, researchers can take advantage of new software tools such as full text search and shared annotations to better find answers to specific questions. But existing digital annotation and media exploration tools do not focus on the research need to understand individual documents within the larger context of an archive. We propose the construction of Spectacles, a digital humanities tool that uses natural language processing techniques to help researchers perform speculative analysis by creating an active archive from any repository of digital texts. Previous tools have focused on helping researchers answer known questions or helping students understand individual texts in a classroom setting. Spectacles, by analyzing both source texts and related annotations, created by researchers in different disciplines using the same repository for different purposes, creates a uniquely useful active archive, a research site that can make recommendations to help direct researchers to questions and answers they had not previously considered. This collaborative, cross-disciplinary contextualization enabled by Spectacles, paired with its recommendation techniques, will aid researchers in performing speculative analysis of digital archives.

I’m excited and optimistic about the potential for computational techniques to aid human understanding of the world. The research I’ll be doing this year will allow me to gain a better understanding of modern natural language processing; but by using NLP to build tools for humanities research and archive exploration, I hope to also gain a better understanding of how humans relate, and have related, to each other and the world.