Transcription of the name field by volunteers is already in progress. We started with names because they are the most useful for finding people, and among the most difficult for automatic handwriting transcription software to read. We will be adding additional fields soon when we are comfortable with the transcription workflow. We will also add an interface for users to submit corrections.
A few fields (name, relation, sex, age, age-at-first-marriage) are already being processed automatically to provide the initial search functionality. Over 80% of the 1930 census is already searchable
due to automatic transcription. Because we started with very few human-transcribed examples to train the system with, the accuracy is noticeably lower than human volunteers can do, particularly on the name field. As more human transcriptions become available to train the system with, automatic transcription will improve.
Many images are already on the server, however not all uploaded images are set up for transcription yet. If you are familiar with census records and would like to help set up images that are already online (especially for your own areas of interest), please contact us
and we will help you get started. For locations that do not have images online yet, you might consider sponsoring
the data for that location– we give higher priority to getting images uploaded and set up for sponsored locations than for unsponsored locations.