17 November 2008
Agenda for Meeting with Robert and Aaron
- Language. What are we unclear about? Let's cement the meaning(s).
- Specifically: percentages of QA - kt: the next 2 shipments (11/20 and 12/4) both IA and SIL will do 100% QA, after that,they will go back to statistical samples.
- What are the shades of "dark"? kt: after finding a level 1 error (see Rob'ts doc) should go totally dark. dark means not indexed.
- Process: Scribe page numbering - how does it work? Why hasn't it?
- Operator responsibility: Page numbers don't add up - who's responsible? kt: operator is responsible for asserting pages and scanning all pages. retraining should have rectified this problem, but there may still be ongoing personelle problems at FedScan.
- IA QA: Why isn't this being caught more often (if at all)?
- What is an "acceptable" number of missing pages (or percentage) in a scanned book? kt: AH, AC, MK will check on contract language, no one knows what the 2% error rate means - 2% of what? Keri and Grace assume it is 2% of books in a shipment, or per a fixed number (something big, like over 250). can NOT be 2% of pages per book!
- Any thoughts to rethink how to deal with missing pages so that entire books need not be re-scanned? kt: no solution in near future, no IA developmer time is devoted to this issue currently.
- Time and $$$: It's cheaper to catch the error while the book is in-hand. Why not take the cheaper solution? kt: new procedure for sil. ia would like us to mix up serials/series and monos on the cart, to help them not grab the wrong book. (agh)
- Foldouts kt: should be in business by end of January
- Will IA consider other foldout solutions (such as are implemented in other scanning operations around the world, Indus, BookEye, SkyView, etc.)
- BHL would like to convene a "QA review of foldouts" to see what we all mean by that
- Can we please get the foldout machine to work at SIL?
- Acceptable errors: what do we mean? (see above under #4)
- Standardize reject rate/process kt: BHL will come up with a written list of standard reject criteria (e.g.,missing t.p. not ok, text cut off on ad pages ok, etc.) for ALL BHL - particular attn. should be paid to gutter, condition, margin, skew and text cut off. a "partner profile" like this will be created by each fedscan customer, in writing. WHY DID NOT ? pin them down to photos of questions and more communications ??
- The Boston Experience:
- Bound withs kt: fedscan says this is OK. slip of paper in book, need barcode for each item.
- Sending snapshots for reject review - kt: maybe. weren't enthused.
- electronic invoices & packing lists - aka, electronic manifests. we should be getting a spreadsheet via email with the manifest of items returning to SIL. when receiving manifests remember to CC Anne Harrison!!
- Chocolates on books when they return - kt: they did not get this joke.
- Demonstration of Wonderfetch (tm) on Stefaan's TSP - kt: did not get to
Misc other things Keri heard and wrote down, and an action item..
- ABYY & luratech questions. MK asked RM to keep him apprised of potential pdf creation packs IA might use to replace luratech. for OCR - they are moving from the server install of ABYY8 to a workstation license for ABYY 9, run from VM windows on a linux box.
- Treasury is starting to scan at FedScan Dec. 1. Dont think they have barcodes.
- Dorothy (night shift mgr) and Quinn are the QA people at FedScan.
- IA statistics and reporting - claim metamgr is dead. want us to use xml thing. Keri will get RM list of fields (from meta.xml) that we all need to search on that aren't in the stupid list now.