Finding Florida—Implementing Machine Aided Indexing in an Academic Library

Abstract

From 2018 to 2021, a team of library professionals at the George A. Smathers Libraries worked to implement Machine Aided Indexing (MAI) in order to locate content about Florida places and spaces among the 16 million pages of hosted content at the University of Florida Digital Collections. This semi-automated process uses a combination of commercial software and locally developed applications. MAI consistently assigns subject terms from controlled vocabularies, aka, taxonomies, to thousands of items in a couple of hours. This method selects terms considering the frequency of the terms appearing in the texts and as well as the preset rules that define the concurrence of terms and other contextual restrictions. After three years’ effort, the team identified 23% items, out of the 76,316 text-based single-volume content in English, are about named places in Florida and tagged all of these items with place names, adding 34,000 access points for these items. Most of these access points were not available prior to this process. On top of that, the team also compiled a Florida specific taxonomy--Thesaurus of Florida Place Names. This paper outlines the key components of this MAI process and details the challenges and lessons learned.