The digitization of art images has created new opportunities for artists, researchers, and students to explore visual arts. However, many digital art libraries are difficult to navigate. This is because they lack metadata to describe their images, which makes finding images based on their content hard. We applied computer vision techniques to a dataset of art images to create an art recommendation system.

UW's Institute for Health Metrics and Evaluation (IHME) publishes papers about disease research to improve population health. They gather feedback from a network of collaborators around the world, and each paper receives up to 5000 comments. We've built a streamlined system for managing comments. Our platform consolidates multiple applications into a central location that utilizes automation and intelligent triaging to process large sets of comment data quickly. IHME staff no longer need to look through thousands of comments and categorize every cell. We have built a magic button that, with just a click, reduces weeks of work into 30 seconds!

The Clark Library serves over 5,000 patrons at the University of Portland. I conducted a collection assessment to understand user behavior and determine areas of growth for ebook resources. I compared usage of ebooks and print books by subject area using ILS data, consortium reports, and vendor reports. The assessment revealed valuable relationships between format preference, users, and subjects, especially in the humanities disciplines. The strengths and weaknesses of each subject area and format will inform collection development decisions for future budget years and will benefit the library patrons by more closely meeting their research demands.

The Public Health Department of Seattle and King County maintains a comprehensive collection of data called the Community Health Indicators (CHI) dataset. These indicators aim to measure health behaviors, health outcomes and demographics of King County residents. The CHI online data portal draws in a broad range of users, including researchers, nonprofits, policy-makers and concerned citizens. However, some users had difficulties navigating and understanding this website. Our team utilized a user-centered design approach to create a three-part solution to tackle these problems. Our solutions include: 1) information reorganization and search functionality, 2) video tutorials, and 3) data visualizations.

We redesigned a tool, called the Broadband Connectivity Assessment, which supports advocacy for high-speed internet access in underserved communities in the US. We conducted user research and interviewed participants who piloted of a beta version of the tool. While responses were overwhelmingly positive, there were challenges which made the tool difficult to use. We offer solutions through a redesign of the user experience and architecture of the tool. Because the needs of participants vary, our final design recommendations focus on creating a tool that is flexible and can be adapted to the needs of any community.

For my capstone project, I processed and created finding aids for five visual materials collections from UW Special Collections. As a result of this project, materials that were previously inaccessible to researchers are now contextualized and fully searchable through Archives West and UW Digital Collections. Multiple steps had to be taken in order to make the materials accessible: researching and arranging materials, housing materials in appropriate archival housings, encoding materials using the XML publishing program XMetaL Author, scanning images and creating digital surrogates, and revising finding aids based on the feedback of UW’s Visual Materials Curator Nicolette Bromberg.

The Prosecuting Attorney’s Office prosecutes all felonies in King County. Their decisions are based on experience but minimal statistics. With increased complexity in crimes and rising need to analyze data from multiple law enforcement agencies, we collaborated with their Firearms Violence department and developed an infrastructure to analyze criminal data. The prosecutors got a better understanding of the trends across key metrics. Now, they can analyze bullet casings to link different incidents, extract an individual’s historical criminal involvement and work with law enforcement agencies to increase referral rate, thereby, improving the sponsor's overall data literacy.

Chronic non-communicable diseases, such as cancer and diabetes, are responsible for over 60% of worldwide deaths annually, and 80% of those occur in developing countries. Yet, only 1% of global health funds are aimed at addressing this group of diseases. According to our sponsor, RTI International, while there is ample data documenting inadequate funding, a lack of effective data visualization inhibits activists from spotlighting this issue. Our project turns data into a captivating set of graphics that tell the story of inadequate funding to reduce chronic non-communicable diseases in developing countries, helping RTI make a stronger case for more funding.

Do you love your work, but have this roadblock called “SQL” you don’t have the time to learn? Our product, DataBot, seamlessly bridges the gap between business and technology, so that you can focus on scaling new heights. DataBot leverages Machine Learning to transform natural language into SQL and presents actionable information. This empowers consultants to retrieve data on the fly, and eliminates ad hoc requests to analysts. DataBot harnesses the power of the cloud to ensure zero downtime. Slack integration ensures it can be easily deployed in an office environment. Embrace yourselves for the next disruptive technology revolution!