Handwritten Text Recognition (HTR)


What is HTR?


Handwritten Text Recognition is technology which aims to deliver full-text search results for handwritten documents. The software behind this technology employs artificial intelligence and probability, which does not rely on transcripts, to produce very effective full-text search results.


Why can't I search the entire product using HTR?


In order to provide the most efficient search possible, HTR is only available at document level. A basic search at site-level or from Search Directories will return hits from the detailed metadata that Adam Matthew Digital's editorial team has assigned to each document. If that term is found in the metadata assigned to a document, and HTR has been applied to that document, HTR software will then search the document for the same result, highlighting any hits. 


Can I see a transcript of the manuscript material?


HTR technology does not produce transcripts. The methodology used takes a different approach to identifying search terms based on artificial intelligence and probability. The result is a marked improvement in the accuracy of highlighted results over transcript-reliant systems.


How are search results displayed in the document?

HTR search results

HTR search results display as yellow highlights on the image within the image viewer. If there are multiple results for the terms or phrase within a document, these display as a list beneath the image viewer. Each result shows a snippet of the original handwritten manuscript with the term or phrase centralised. Click the snippet to open the image.

HTR snippets


Can I use Boolean operators like AND or OR?


Boolean operators are supported in an advanced search. The HTR software searches for the entered search term exactly, so a search for women AND employment will search for all three words together, not ‘women’ and ‘employment’ as two terms occurring separately in the same document.