What lies beyond the recognition wall : open-source tools for HTR post-processing and publication in digital scholarly editing

Recent progress in Handwritten Text Recognition (HTR) has enabled many researchers to speed up the process of transcribing their large collections of manuscripts. This dissertation is an endeavor to tackle what comes after this transcription stage, where annotations need to be made and the product of this labor needs to be published. It uses the well-established FAIR principles and the maxims of Open Source as sturdy foundations to construct a methodology for developing tools not only for digital scholarly editing, but for research software in the broader sense. As its dual case studies and proposed solutions to the issues of HTR post-processing, this research chronicles the development of two open-source research software tools and their use cases. These tools are Axolotl, a collaborative TEI-XML editor, and Necturus, a TEI-XML publication platform with a plug-and-play template variant. These tools both sustain the connection between the scanned image of the manuscript and the extracted text, a connection which is a by-product of using HTR platforms for transcription. They have been deployed in real-world projects ranging from individual scholarly editions to institutional collections of 90,000+ manuscript pages. Axolotl and Necturus attempt to build up a sustainable pipeline for researchers to post-process HTR output by providing solutions for annotation and publication, but they are also emblematic of the principle that research software should be treated not as an accompaniment to research, but as research in its own right.