Current research on designing digital library search interfaces for public library patrons is limited. Historically, research on exploratory search has tended to focus on distinct populations such as academics, professionals, and students, making it more difficult to extend these findings to the general public where the information needs can be broader and may be ill-defined. To help address this gap, we initiated a collaboration between academic researchers and librarians at our university, and established partnerships with the Regina Public Library (RPL) and its library technology service provider – Saskatchewan Information and Library Services Consortium (SILS). We were able to secure funding from the Natural Sciences and Engineering Research Council (NSERC) through its Alliance Grant program.

Research Collaboration & Partnership

Our research to date has focused on two interrelated objectives: 1) understanding the strategies and processes that public library patrons employ when undertaking complex search scenarios, and 2) designing, developing, and studying search interface prototypes that support complex search tasks and exploratory search processes.

For the first objective, we used retrospective think-aloud protocol and thematic analysis to understand the behaviours of public digital library searchers when undertaking complex search tasks. We learned that while the cognitive processes described by the participants are aligned with the exploratory search model (e.g., query reformulation and exploratory browsing to learn more about their topics and address fluctuating uncertainty), other processes were also employed. In particular, the participants also demonstrated meta-cognitive processes, such as planning their search, and strategic decision-making based on pragmatic factors such as time, which were not captured in the exploratory search model. We are currently replicating this study with librarians conducting the search, so that we can compare the differences between the search strategies employed by library patrons in comparison to expert librarians. Our ultimate goal in this line of research is to provide guidance on how future search interfaces might be designed to enable library patrons to search like librarians.

In pursuit of the second objective, we have created and conducted user studies on a variety of public digital library search interface prototypes, each of which addresses a difficulty that searchers face when undertaking exploratory search processes. Examples include a drag-and-drop query refinement approach to assist searcher to interactively revise their queries when using their mobile devices; a timeline visualization of exploratory search activities to enable cross-session searching; and a nested workspace approach to support searchers in organizing found information. Other work in progress includes using the timeline visualization as an anchor for providing query refinement suggestions, and studying visualization approaches to support searchers with dyslexia to see patterns among search results, minimizing how much they need to read to find relevant resources as they pursue complex search tasks.

The pursuit of this research agenda was made possible through the collaboration and partnerships we established. Discussions between academic researchers and librarians has helped us to co-create the research objectives and ensure that the pursuit of these objectives remains grounded in the needs of public library patrons. The partnerships with RPL and SILS have enabled the recruitment of participants in our pursuit of the first objective, and API-level access to the public library search infrastructure used in actual public libraries. We encourage others seeking to pursue research on public digital library search to establish similar partnerships.

Thoughts for the Future

As these projects come to fruition, we set our sights on what is next. As Large Language Models (LLMs) and Generative AI (GenAI) change the nature of searching the web from that of seeking resources to that of seeking answers generated from top-ranked resources, it is important to consider how this paradigm shift might affect online searching within public libraries. Important questions to consider are to what extent can question answering address public library search activities, and what other aspects of public library search can be enhanced with LLMs and GenAI. The first generation of LLM-enhanced search has focused on how to replace the human effort of evaluating search results with the automatic generation of answers. We are on the cusp of a second generation of LLM-enhanced search, where the focus is on augmenting other elements of the human effort associated with searching when question-answering is insufficient. For example, when undertaking an exploratory search, LLMs can be used to automatically summarize saved resources. Approaches such as this do not seek to turn every search problem into a question answering problem; instead, they consider the needs of the public library patrons and seek to enhance how they search rather than change the nature of searching. The continued study of how to augment the human element of exploratory search within public digital libraries using the latest LLM technology will require continued collaboration between academic researchers and librarians, along with partnerships with public libraries and library technology service providers.