Comments on: Digital Tools for Television Historiography, Part I

By: Darragh McCurragh

Darragh McCurragh — Fri, 12 Jun 2015 21:02:51 +0000

For thousands of years, dating back to before the library of Alexandria, recording was done in writing and on paper, leather or parchment. Only since a few decades do we “trust” magnetic and optical media with our information. However, recording formats as well as technology change decade after decade after decade. There already are lots of “vernacular” media that even specialists cannot reliably read. To really talk in terms of historical time frames, like centuries or millenia, we urgently need a constant (!) perpetual migration strategy for each media type and recording format to “copy” it onto the next new technology and then the next, the next and so on. It already seems foreseeable that libraries will have more work to do with the “immaterial” storage technologies than they once had cataloging and storing books!

By: Andrew Bottomley

Andrew Bottomley — Wed, 03 Jun 2015 15:07:16 +0000

Elana and Cynthia, this is a great discussion. Thanks!! This Macworld article, although a few years old now, provides a helpful analysis of the file size and accuracy of different OCR software. Author’s conclusion is that Acrobat Pro is the least accurate. Best configuration seems to be ABBYY/DevonThink Pro Office scanning at 300dpi in grayscale at medium compression: http://www.macworld.com/article/2043857/secrets-of-the-paperless-office-optimizing-ocr.html

By: Cynthia Meyers

Cynthia Meyers — Wed, 03 Jun 2015 02:10:53 +0000

Yes, this retrieval issue is what concerns me about digital search. If the document doesn’t come up–whether because of bad OCR or a bad search term–I might not remember the document exists! So why collect it if I can’t retrieve it? I am taking more notes in DTP than I expected precisely to create searchable text for the “bad OCR” documents–and even the “good OCR” ones!

For better OCR in DTP, be sure to go to Preferences>OCR and select “accurate.” Slower but, presumably, more accurate!

By: Elana Levine

Elana Levine — Wed, 03 Jun 2015 01:43:47 +0000

Cynthia, I love this idea about the Bad OCR label! Just today I was looking for an article I knew I had but it wasn’t turning up in my search. I eventually found it, but it made me think about what I might be missing because of my (pretty common) Bad OCR documents. I think I will start using a system similar to yours as I read through my next batch of materials. Good to know DTP is doing OCR better than Acrobat; I use both but will tend toward the former now (even though I don’t plan to collect too many more sources!).

By: Cynthia Meyers

Cynthia Meyers — Wed, 03 Jun 2015 00:44:46 +0000

Great post–thank you! Following your recommendation, I am now also beginning to use DevonThinkPro Office. I use the version that includes ABBYY reader software, which is better OCR software than Adobe Acrobat (Acrobat refuses to OCR my PDFs created from iPhone photos). In DTP, File>Import>Images with OCR lets me import files into DTP and OCR them in large batches–though not quickly, because I use the “accurate” OCR setting.

After importing I can click on the “Word” icon and see immediately in a sidebar which words were found by the OCR. This is key! Some old carbon copies just don’t OCR well (and the word list is full of crazy nonwords), which means a later search might not locate the document. So I invented a label (an orange dot) that means “Bad OCR,” and I annotate that document in “Spotlight comments” with more searchable terms and phrases so that it will be found in a later search. I hope that labeling some documents “Bad OCR” will also help me identify which documents might need double checking if a search doesn’t find them.

Thanks, Elana, for helping us index card junkies go digital!