Trial: Colonial America, with handwritten text recognition

Adam Matthew Digital (AMD) has completed three of five modules of Colonial America, an online resource that will include all 1,450 volumes of the CO 5 series from The National Archives, UK, covering the period 1606 to 1822. The Library currently has trial access to the three modules until October 16, 2017.

It is with the third module of Colonial America that AMD has implemented a technology that allows for full text searching of handwritten documents. The Handwritten Text Recognition (HTR) application uses algorithms and artificial intelligence to determine possible combinations of characters in manuscripts.

The default search will search both metadata applied to documents and their text. When results are found in the text, they are displayed as snippets.

example of search results

 

 

 

 

 

Clicking on a hit will take you to the page where the word appears.

example of page

 

 

 

 

 

 

 

 

This search function is ground-breaking, but not 100% accurate. I’ve searched for words that exist in a document and have retrieved no results. I have also searched for words that were written sloppily or with a long s and have retrieved results.

I am interested in your feedback on both the value of the database and your successes (or failures) with full-text searches. Email me at dorner@berkeley.edu.

(Please note that PDF downloads are not available during the trial.)