This capstone project wanted to determine characteristics of materials shared internationally by libraries’ interlibrary loan (ILL) departments, seeking to determine drivers behind such resource sharing and whether any interesting or useful patterns in the practice could be identified. By interviewing institutions who participate in international ILL, we found that the experience, and number of resources shared by the participants varied greatly and were the result of differing circumstances. It was determined that participants faced similar challenges the world over, but that the responsibility ILL teams feel to their users to provide the best access to resources is universal.

The HEAL-WA team aims to increase the use of their portal by clinical care providers, meeting their mission to “provide evidence-based information to support patient care.” I created a Care Provider Toolkit to increase access to databases and journals useful to clinicians. The toolkit's clear and easy-to-use design encourages the use of these resources. With reliable access to evidence-based information clinical care providers can provide quality care to patients. Providing evidence-based resources at no additional cost creates options for healthcare workers who may otherwise be limited in their ability to use these resources due to the high cost of access.

As a federal agency, the National Oceanic and Atmospheric Administration (NOAA) must comply with Executive Order 13985, which calls for federal agencies to address the lack of diversity in the federal workforce. To meet its diversity objectives, NOAA must engage with underrepresented student populations for internships and job opportunities. Our recruiting tool assists Washington and California state NOAA offices in their recruiting efforts, with the hopes of expanding to more U.S. states in the future. By leveraging public data, the tool provides a straightforward way for recruiting staff to find schools serving underserved communities.

Large Language Models (LLMs) attempt to imitate human behavior by responding to humans in a way that pleases them, including by adhering to their values. Therefore, it is important to understand whether LLMs, upon understanding a user's national background, will showcase a different set of values to the user. We have found that LLMs have an innate understanding that cultures differ in values, responding differently based on different nationality input. By discovering this information, we empower others to consider whether using LLMs to seek advice will be ethical and beneficial for them, especially with answers that must be culturally-sensitive.

With the rise of AI-powered language models, there has been growing interest in their creative and critical capabilities. In this study, we investigated ChatGPT generating and evaluating poetry. We conducted experiments to test ChatGPT's proficiency in categorizing and producing poems in various forms and styles. Our findings reveal distinctive linguistic patterns in ChatGPT-generated poetry, such as the prominent use of the first-person plural (“we,” “us,” “our”), as well as forms and styles that the model can identify with surprising accuracy. Our research contributes to a growing body of literature on AI-powered language models in the field of literature and poetry.

In our capstone project, we are converting Maine community health assessment reports and other health data into dynamic visualizations on a Power BI dashboard for the general public. Our main goal is to highlight the geographic and economic disparities in public health access across Maine. Although our project centers on Maine, the findings could be relevant to anyone dealing with public health issues in similar states. This effort not only informs the community about local health challenges but also equips them with the information needed to advocate for improved health policies and practices.

For this capstone project, we explored a collection development challenge for Washington State's open data program. Specifically, we investigated data selection practices at leading government open data programs in the United States. “Selection” in this context refers to the range of practices that determine what and how much data is opened and ultimately published on open data portals. Our final report, which summarizes the state of data selection and best practices across the United States, will help Washington State Library connect individuals and communities to data that will help them flourish.

In collaboration with Self Help Graphics and Art (SHG), this project processed, condition reported, created metadata, and rehoused 315 archival artworks by the organization’s co-founders, Sister Karen Boccalero, Antonio Ibañez, and Carlos Bueno. The research culminated in two finding aids for each sub-collection. These collections evidence the origins and early artwork of the co-founders, who collectively lit the fire for an organization with 50 years of history and significance to Chicanx and Latinx artists in Los Angeles still burning bright today. The project aims to make these previously unprocessed collections accessible to SHG’s local community of neighbors, artists, and researchers.

Our project seeks to augment the existing Blacklight tool to enable users to compare current and previous ‘scans’ for a website, so that users are cognizant of how companies manage their data and are better equipped to take steps to secure it. The Blacklight Archiver, a product of tech publication The Markup, enables users to easily track online surveillance. By providing users with point-in-time comparisons of cookie-based website trackers, Blacklight users are able to see which websites are the largest surveillance offenders. This publicly-available product is useful to journalists and consumers alike.