Migration: It isn’t just for the Birds
Post on 31-Jan-2016
54 Views
Preview:
DESCRIPTION
Transcript
Migration: It isn’t just for the Birds
Presented by Becky Bell MPLA/NDLA/SDLA
Tri-ConferenceOctober 5, 2002
Data Migration What is it? Who does it? How can you prepare? When is it done?
Data Mapping MARC Records Non-MARC Records
Migration ConsiderationsData MappingSerials Migration ComponentsData Mapping Detail
MARC records Non-MARC records
Data Cleanup
What is data migration? Simply, moving data existing in one software
application to a different software application Usually is more complex than expected due to
Changed functionalityField size incompatibilityDisparate record structuresExisting/new/no standards supported in either
previous or successor software
Two “Truths” about Data Migration 1. There is always a bigger mess than you
thought. 2. Your assumptions are always wrong.
• From “An ABC guide to data migration,” by Ross Bentley, Computer Weekly, Feb 7, 2002, p.39, and attributed to Richard John, managing director of data migration at Alchemy.
“Data migration not the most fascinating aspect of implementing a new system.”
Data migration is a matter of ABC Analysis of data Building conversion programs Cleaning incorrect data
Remember GIGO--Garbage in, garbage out
Who does data migration? Any entity (library, library consortium,
banks, retail stores, etc.) moving data from one software application to a different application
Focus today is library data Ideally, library staff and vendor work
together to create data mapping specifications
How can you prepare?Read and learn about
• Windows • MARC21 Holdings Format
Know what you currently have• Become familiar with your system’s data dictionary• Document your workflow• Document local practices
Work on data cleanupLearn as much about your new system as possibleTalk to others who have completed their migration
When is it done? Obviously, the switch to production Milestones along the way
Decisions made about what must migrateData mapping specifications writtenSubset(s) converted, loaded, and indexedData reviewed for mapping and data accuracyAs many iterations as needed to get what you
expect
Keep in mind that if data migration is skimped, muffed or even rushed, the success of the new system can be dangerously compromised.
Details matter!
Migration Considerations
What data must be migrated?What data are you willing to re-key?What data are you willing to lose?What data are you willing to share if
you’re part of a consortium?Can uncataloged/unlinked items
exist in your new system?
Migration Considerations
Can bibliographic records exist without items and/or holdings records?
How much training and/or consulting services from the vendor can you afford? Pre-conversion training? Systems administration? Functional training?
Remember that policies and privileges aren’t migrated; you’ll need to create them in new system.
Serials Migration Components
StandardsMARC21 Bibliographic FormatMARC21 Authority FormatMARC21 Holdings FormatNISO Z39.71-1999 Holdings Statements for
Bibliographic Items
No standards forSubscription, claiming, routing dataAcquisitions data (Orders, Vendors, Budgets)
Serials Migration Components
Serials Serial data will be stored in bibliographic,
holdings, item, acquisitions, and serials modulesWon’t look at Patrons, ILL, Booking mapping today
Linkages to maintain or create: serial record to bibliographic and holdings records to items and orders
Serials records not used for check-in can be mapped to a bibliographic record (and can be suppressed from the public catalog)
Data Mapping
Two types of library records MARC
BibliographicAuthorityHoldings
Non-MARCItem
Data mapping
Patrons• Loans• Holds• Cash Transactions• Blocks
VendorsOrders
• Monograph• Serials• Standing orders
Purchase orders (invoices)
Data mapping
BudgetsSerials
• Subscription records• Publication patterns• Claims• Routing lists• Binding records
Interlibrary Loan recordsBooking records
Data mapping detail
MARC bibliographic records If consortium, need to choose system
architecture first (if there are options)One bibliographic/authority/holdings database
for all or separate databases for each library?If one bibliographic database, migrate each
library’s record or merge like records?If records merged, what are the merge criteria?Physical union catalog or virtual union catalog?
Data Mapping Detail
Examples of differences in bibliographic records for Cricket and Library JournalCricket cataloged by Jamestown College
(NDJ) on January 24, 1991 and by Fargo Public Library (NFG) on February 1, 2000
• Differences– NFG has a revision date in 010– NFG has additional subfields in 040 ($$d NST
$$d NSD $$d DLC $$d IUL)– Tags 260, 300, and 321 formatted slightly
differently, probably based on cataloging rules current when cataloged
Data Mapping Detail
– 362 tag formatted pre-Z39.71-1999 Holdings Statements for Bibliographic Items in NDJ and post-standard in NFG
– 901 tag capitalization differs– 936 tag identifies different surrogates used
for cataloging• Additions
– NDJ record has 265 and 350 tags included– NDJ added local holdings information in 500
tag– NFG record has 037 and 130 tags included– NFG record has two 500 tag notes
Data Mapping Detail
• NFG added a 590 tag for local holdings information
• NFG includes 850, 890, and 994 tags
Data Mapping Detail
Library Journal cataloged by Augustana College (SDA) on May 2, 1991 and by Dakota Wesleyan University (SDW) on November 5, 1992
• Differences– In 012, SDA has $$i 8510 and $$m l– In 012, SDW has no $$I and has a $$m c– In 040, SDA has many more $$d’s than SDW– Contents of 265 different—which is more current?– In 321, SDA includes a $$b– Contents of 350 different—which is more current?– Second 650 in SDA record has a $$w
Data Mapping Detail
– 780 in SDA record has a $$w– Contents of 936 are different in SDA and SDW
• Additions– SDA record added a 310, a 362, nine 510
fields• What happens when the newest record overlays
the first record loaded?
Data Mapping Detail
Possible merge criteria Each record converted to new system
Pro--each library retains its cataloging and retains local cataloging policies
Con--lots of duplicates
First bibliographic record into system becomes record for all, based on one common field (bib id number)Pro--cataloging may not be acceptable to allCon--Keeps database size smaller
Data Mapping Detail
First bibliographic record is supplemented as additional records are convertedPro--Additional access points get added for
retrievalCon--Records can become very long Con--False search results become more likely
for your patronsCon--Improper MARC tags add information
more appropriately stored in the holdings record
Data Mapping Detail
Merging algorithm determined, for example, 100, 245 subfield a, 250, 260 subfield b and c match exactly
Pro--De-duplication of database accomplishedPro--All like titles merged into one recordCon--Like titles may be different formats
Sample specifications from PALS to Aleph mapping
No one right way--study options, then choose what provides results you desire
Data Mapping Detail
MARC Holdings records Legacy library systems won’t have holdings
records to convert, so records will be created Specifications need to exist that detail what
data goes into tags and subfieldsNUC symbol, location, call number, and pre- and
post-input stamp information Serials captions Enumeration and chronologyURL stored in 856
Data Mapping Detail
Local notes
Will need to create location/call number mapping documentSample location mapCall number pick listCall number (x) suppression099 call numbers or other non-standard
numbersSuDocs numbers
Sample specifications
Data Mapping Detail
Items Uncataloged/unlinked items will need
bibliographic records Specifications need to include mapping
bibliographic data in items to proper MARC fields for brief records
Links between subscription records and items need to be defined
Data Mapping Detail
Sample itemsSample specificationsSample data map
Data Mapping Detail
Course reserves If serials titles or issues are on reserve,
need to preserve reserve information and status
Sample specificationsSample data map
Data Mapping Detail
Vendors Used by both acquisitions and serials
Acquisitions uses for ordering and payingSerials uses for claiming missing issues
Sample specifications Sample data map
Data Mapping Detail
Orders Orders need to exist for active serials
(check-ins) Sample specifications Sample data map
Data Mapping Detail
Purchase orders (invoices) Payment history Sample specifications Sample data map
Data Mapping Detail
Serials Prediction patterns
Library Journal exampleCricket exampleCoding detail
• Seasons• Free text• Supplements or other additions to regular pattern
Memberships
Data Mapping Detail
Inactive serials records Decision/internal/history/public records
Sample serial recordsSample specificationsSample data map
Data Clean-up
Bibliographic/Holdings records Identify commonly misspelled words in records
in your catalog (use http://www.aallnet.org/sis/tssis/tsll/22-03/obs-locl.htm as a source) and correct
Identify records with duplicate call numbers Identify records with no call numbers Browse your title index for incorrect filing
characters (A, An, The) and correct Begin discussions about indexing
Data Clean-up
Non-MARC records Review reports for obvious errors Weed collections Resolve outstanding overdues, fines and
bills Inventory Review existing notes and remove those
no longer needed Monitor binding lists
Data Clean-up
Purge closed orders and purchase orders (invoices)
Review open orders
Recommendations
Staff development issues Windows MARC21 Holdings Format training Update (or create) procedure manuals to reflect
current practices Document local practices Attend conference programs about standards
and migration issues
Subscribe to open mailing lists for next vendor
Recommendations
Involve all staff in data mapping planning and data review
Develop a good sense of humor--it will be one of your most valuable tools
Remember you have a life outside of data migration!
In Closing,
Review your dataWrite detailed conversion
specificationsReview converted data by comparing
data in old and new systemsDetails matter!
top related