Finding ways to enhance searches for users seeking to find African American literary texts across variations in spelling, naming conventions, and incomplete information remains a central challenge and opportunity.
Fuzzy searches constitute one important component of our efforts with the Literary Navigator Device to improve discoverability and return more open, user-responsive results. Fuzzy searches are queries that allow for approximate matches rather than exact ones, accounting for misspellings, alternate spellings, and partial inputs. In other words, a fuzzy search increases the likelihood that users will locate relevant materials even when their search terms are imprecise or incomplete.
This technique is usually left to computer scientists and librarians, not scholars of African American literary studies. Generally speaking, we are not involved in the construction of tools and large databases used for information retrieval, so we do not always think about how search systems shape what users can find. But perhaps our expertise might contribute to refining these tools in ways that better reflect the complexities of African American literary production.
Over the last several months, I’ve spent more time thinking about fuzzy searches more than ever before because of the regular meetings I have with Meg Smith, director of our university DH Center, and Dan Schreiber, the web developer for the center. They provide implementation and consulting for our overall site Black Lit Network, and their work on the Navigator has been particularly useful and illuminating.
Early on, during our meetings, Meg and Dan would talk through the workings of the search feature. Before coming to firm conclusions about what would work best, they were thinking carefully about the different kinds of searches various users might pursue.
Right now, we have configured the Navigator so that a wider range of results appears in response to user queries. For example, when someone searches for Toni Morrison, they will see her novels, but they may also encounter works by Brittney Morris due to the shared “Morris” within “Morrison.” Similarly, results may include works by Terah Shelton Harris because of overlapping letter sequences, as well as authors like Teri Woods, where the system registers similarities between short first names such as “Toni” and “Teri.” In effect, the fuzzy match settings are designed to cast a wider net, though that breadth can sometimes produce unexpected or only loosely related results.
We will continue refining these settings, adjusting sensitivity levels, weighting exact matches more heavily, and pinpointing how closely terms must align. We'll keep doing that until we arrive at settings that best serve users searching for African American literary texts. The goal is to balance openness with precision, while also ensuring there is room for unexpected discoveries.
Related:
No comments:
Post a Comment