Year

2017–2023

Role

Co-designer with other librarians

Contributions

Designed for transparency about the technology used and information accuracy in edge cases; Implemented, maintained, and refined CSS and HTML customizations to improve user experience

Single-Sentence Summary

My librarian colleagues and I created a natural language processing-based chatbot, ensuring patrons could receive reliably-true answers to frequently-asked questions even when no actual humans were available for chat.

Intentionally simple line drawing of a cute robot smiling and holding up its arms like a weightlifters, with the words How Can I Help written on its torso.
Library chatbot persona, hand-drawn to reinforce the tool's simplicity.

Project Details

Brief History

In 2017, my librarian colleagues and I created a chatbot to answer library patron questions when no actual library staff were able to answer questions ourselves. Although community college students find it most convenient to work on assignments or do research over the weekend, the library’s budget meant that we had reduced Saturday hours and zero Sunday hours.

Using the natural language processing capabilities of the IBM Watson service on Bluemix / IBM Cloud, we were able to effectively create a “reverse phone tree”-style chatbot. You could imagine this as an FAQ where the chat window led the user to the correct answer, rather than to a generated answer. When patrons asked questions, the software identified keywords and responded with follow-up questions that directed them toward appropriate answers.

Kudos

Did you enjoy this? Let me know: