OCR Functionality in ILLiad

OCR (Optical Character Recognition) is a feature of text documents that allows the visible text to be turned into searchable data and/or parsed by screen readers. This allows a user to use the search function on a PDF (control + F) to search for specific terms in the document. The transmission of a file via ILLiad/Odyssey does not affect OCR within a document. However, other circumstances may affect whether or not a specific document supports OCR.

OCR Compatibility

The transmission of a file via ILLiad does not itself do anything to OCR within a document. It is also important to note that how the Lending institution generates the file is equally important as the behavior of the Borrowing library. The following chart shows a breakdown of possible processing methods and their effect on OCR compatibility:

OCR Works?	Processing Method
No*	Lending library scans the document via ILLiad, then sends a TIFF via Odyssey or Article Exchange
No*	Lending library scans the document via ILLiad, exports the file and re-imports into ILLiad as a PDF, then sends PDF via Odyssey or Article Exchange
Yes	Borrowing library adds an OCR-compatible cover sheet to a received OCR'd PDF document.
Yes	Lending library imports a native PDF file, sends via Odyssey or Article Exchange, and the Borrowing library does not add a cover sheet

*ILLiad 10: an image-only document received by a site that has the Article Exchange OCR handoff enabled can be made searchable automatically after delivery — see Automatic OCR via Article Exchange below.

In summary:

• If the file was ever a TIFF, or if the Borrowing library appended a cover sheet to the PDF - OCR will not be functional; Unless the cover sheet was OCR compatible and the OCR was not lost after the PDF file and the OCR cover sheet was merged- OCR will be functional. In ILLiad 10, an image-only document can be made searchable through the Article Exchange OCR handoff (see below).
• If the file was a native PDF and sent via Odyssey or Article Exchange and no cover sheet was added - OCR will be functional.

Cover sheets can affect OCR, as shown in the table above. For how Odyssey Manager merges a cover sheet and what it preserves in ILLiad 10, see Odyssey Coversheets.

info

TIFF documents, due to how text compression occurs, do not contain an OCR text layer. In ILLiad 10, an image-only document can still be made searchable after delivery through the Article Exchange OCR handoff described below, when that handoff is enabled.

Automatic OCR via Article Exchange (ILLiad 10)

ILLiad 10 can make an image-only document searchable automatically by handing it off to OCLC's Article Exchange OCR pipeline. When a received document is an image-only PDF (no selectable text) and the receiving (borrowing) site has the handoff enabled, Odyssey Manager uploads the document to OCLC to be OCR'd and finishes delivery once the searchable version comes back.

• The handoff is enabled per receiving site by the ArticleExchangeOCREnabled customization key.
• It fires only when that key is enabled and the document has at least one image-only page. A document that is already fully searchable is delivered as-is.
• While the document is out for OCR, the transaction sits at the Awaiting Article Exchange OCR status; delivery completes when the OCR'd version returns and replaces the delivered file.
• The handoff is best-effort: if OCR cannot run (for example, an upload error or missing OCLC Article Exchange credentials), the original, non-OCR'd document is delivered instead — the delivery is never blocked.

This handoff is handled by the .NET Odyssey Manager.

Customization Manager settings

The handoff is controlled by the following ILLiad customization keys:

Key	Location	Default	Purpose
ArticleExchangeOCREnabled	System | Article Exchange (read per receiving site)	No (off)	Turns the Article Exchange OCR handoff on. One of the two conditions that trigger OCR (the other is that the document has at least one image-only page).
OCLCWebServiceV2APIKey	System | OCLC (read at parent site)	(empty)	OCLC Web Service v2 key that authenticates the Article Exchange upload.
OCLCWebServiceV2APISecret	System | OCLC (read at parent site)	(empty)	Secret paired with the key.
OCLCILLAuthorization	System | OCLC (read at parent site)	(empty)	OCLC ILL authorization number for the symbol.
OCLCILLPassword	System | OCLC (read at parent site)	(empty)	Password paired with the authorization number.

All four OCLC Article Exchange credentials must be set. If any is empty, the upload fails authentication and the document is delivered without OCR — the delivery still completes.

Document language and OCR language (ILLiad 10)

When a lending library sends a document via Odyssey in ILLiad 10, the document's language travels with it in a new atlas.document.language header. The borrowing library uses this to select the correct OCR language automatically instead of guessing.

• If the borrowing library already has a language on the transaction, the lender's value takes precedence, since the lending library knows the actual language of the document it is sending.
• The lending side's language comes from the transaction's Language field (a multi-select field new in ILLiad 10 that can also be populated automatically from a PubMed citation).
• This is backward-compatible: libraries still on older ILLiad versions ignore the header, and delivery works normally in both directions.