,
13 tweets,
6 min read
Read on Twitter
[01] Good afternoon. My name is Philip R. Burns, better known as Pib. I am a developer in Research Computing Services at Northwestern University. I’ve been here since 1974. Today I am going to speak about the Earlyprint Library project. #HCTwitterConf19
[02] EarlyPrint Library aims to create a free public deduplicated digital library of English books published before 1700. Each entry aims to be a digital “combo” with a good transcription, quality page images, and consistent bibliographic and word-level metadata. #HCTwitterConf19
[03] The current texts are enriched and partially corrected versions of works transcribed by the Text Creation Partnership (TCP) and released into the public domain. We created custom Text Encoding Initiative (TEI) versions from the SGML originals. #HCTwitterConf19
[04] Each text is tokenized and morphologically adorned. Each word has a permanent ID, original and modernized spellings, a lemma, a part of speech, and other attributes. Work level metadata comes from several sources including the Early Short Title Catalog. #HCTwitterConf19
[05] The display for a work features a side-by-side formatted page of text on the left and a page image (when available) on the right. Multiple search criteria for texts are provided (see texts.earlyprint.org/works/). #HCTwitterConf19
[06] The Library offers multiple search criteria to find texts in the collection. See texts.earlyprint.org/browse for the current set. #HCTwitterConf19
[07] The Library also features a framework to support collaborative curation. Users can correct common types of textual corruption in the transcriptions one defect or page at a time. We will be improving the correction framework over time. #HCTwitterConf19
[08] We are also experimenting with automatic correction based on machine learning methods. Automatic corrections can be quickly reviewed and accepted or corrected by human reviewers. #HCTwitterConf19
[09] We grade the texts based on counts of known defects. "A" texts (25%) exhibit no known defects. "B" texts (25%) have a few defects. "C" (25%), "D" (15%), and "F" (10%) texts exhibit increasing numbers of defects. We regrade texts as defects are reduced. #HCTwitterConf19
[10] Project members include faculty, staff, and students at Northwestern University, Washington University in St. Louis, and (in the previous iteration) Notre Dame University. Individuals from these and other institutions contribute text corrections. #HCTwitterConf19
[11] In June 2019 we received a grant from the Mellon Foundation to continue development of the EarlyPrint Library. You may view the current in-progress version of the EarlyPrint Lab site at texts.earlyprint.org. #HCTwitterConf19
[12] If you have trouble accessing EarlyPrint Lab at texts.earlyprint.org, please try the alternate backup server at devtexts.earlyprint.org . #HCTwitterConf19
[13] Thank you for your interest and attention. I welcome questions and comments about EarlyPrint Lab. #HCTwitterConf19
