An interesting short article about plans to combine automation and crowd sourcing on the 3.25 million images of the US 1940 census which is being released next April. The article describes some of the challenges in automating transcription of the census, using 1930 as a test-bed, and plans for image-based information retrieval using supercomputer processing to avoid the costly transcription process. Crowdsourcing would be used to improve the automated transcription.
Read the article at:
No comments:
Post a Comment