About me

Welcome! I’m Anna Martin-Boyle, a second-year Ph.D. student in the Department of Computer Science and Engineering at the University of Minnesota, Minneapolis, advised by Dr. Dongyeop Kang of the Minnesota NLP Lab. My research interest is in the development of natural language processing tools for enhancing digital libraries and scholarly writing interfaces, in order to improve the discoverability of scientific research.

The exponential growth of scholarly publications poses a significant challenge for researchers who must navigate the vast and rapidly expanding landscape of academic literature. As the quantity and complexity of information increase, researchers face cognitive overload and fatigue, struggling to stay abreast of the latest advancements in their fields, synthesize relevant findings, and identify research gaps. The time-consuming nature of literature discovery, reading, and writing scholarly documents further exacerbates these challenges and can hinder the overall efficiency of the research cycle.

My prior work has focused on the reading aspect of the research cycle. During my master’s program, I created a dataset and designed a system to recognize task descriptions in computer science papers by comparing various traditional sentence classification methods and BERT-like models. In the first year of my PhD, I completed a project focused on extracting definitions of mathematical symbols from complex sentence structures. The implementation of this technology aims to improve scholarly reading interfaces by enabling readers to effortlessly access information units such as symbol definitions.

More recently, my research has concentrated on scholarly writing by developing a taxonomy of scholarly writing actions and intentions to better understand the scientific writing processes, based on the Cognitive Process Theory by Flower and Hayes (1981). Our next step is to conduct a large-scale human research study to analyze collected keystroke data and gain insights into the scholarly writing process. If you are interested in participating in this study, please visit our recruitment page.

I am also interested in integrating the scientific literature discovery process with other aspects of the research cycle. Currently, we are exploring how large language models (LLMs) can be used to filter and group sets of scholarly documents in collaboration with human researchers, to aid the literature review process. In particular, we are exploring the Human-Computer Interaction (HCI) side of this problem, seeking to understand how to design systems that encourage and amplify the traits we most value in researchers. For my future work, I hope to address the question of whether simplifications generated by LLMs act as thought-terminating heuristics by instilling a false sense of clarity (C. Thi Nguyen).