Ted Schwitzner and Anita Foster
Illinois State University
Agenda:
Ted - Rationale for the project; getting data out of Voyager
Anita - Getting Voyager data into Verde
Ted and Anita - Recommendations for attendees
Undertook this project because had a need to track eresources in a system with more eresources functions than Voyager - hoping attendees can learn from their successes and unexpected glitches.
Because Illinois State is a CARLI member, their Voyager is hosted by the consortium; Verde is hosted by Ex Libris.
Ability of Verde to look up PO and invoice data in Voyager requires:
--Web Services module installed for Voyager (requires a newer version of Voyager than they had at the time)
--Voyager Line Item ID number entered in Verde Acquisitions record PO Number field
They encountered data and display issues when got this lookup working. Some issues were caused by their local acquisition and business practices:
--Multiple invoices and credits per title per year in Voyager
--Display not uniform between browsers (found that Firefox worked well; IE not so much; sometimes encountered version problems with Firefox updates)
--Display of invoice data not user-friendly in Verde - display of multiple years of invoices/POs was a sort of horizontal paragraph form that IE users couldn't see all data; FF users had to resize windows.
All these issues led them to want to migrate Voyager Acq. data into Verde.
Voyager-Formatted data:
--Fields with set length, required contents - contained some data of interest
--Voyager free-text data - most electronic resource decisions, history, and EDI data were stored in purchase order notes - since this was most data of interest, had a hard time parsing out data into something usable.
Decided to load the following into Verde: purchase orders, line items, and invoice line items from previous two years (FY08 and FY09)
Also decided to limit to only continuations
Excluded "completed" purchase orders - had something to do with ending print subscriptions (a local practice), but this came back to bite them.
Skipped invoice line item notes - EDI data formatted but complex; EDI data not consistent by vendor.
Data extraction from Voyager:
Required 4 subqueries to extract data; resulted in 46 fields and over 9000 records that weren't quite load-ready.
Anita:
Fields in Verde:
--Found it's easier to determine subscription dates in Verde, but there's no place to do that in Voyager; assumed that most dates were calendar year and hoped to find others through trial/error.
--Verde fields used:
Start/end date
Acq status
Vendor
Advance Notification period including auto-renewal
PO Number (Voyager Line Item number)
Material and subscription type
Final price
Other info entered into Verde:
--Any titles included in a subscription
--Number of concurrent users
--Notes such as subscriptions accessible by user name/password only (they don't activate these, but wanted notes so wouldn't constantly be revisiting these titles).
Locally-defined fields:
--Acq status: title change, transfer to new interface
--Material type: Streaming Media
--Subscription Type: Maintenance fee
--Subscription Type: Membership
Loaded FY 08 to current data to build a context for new and renewed acquisitions and to see price over time. Also wanted a solid basis of data for generating reports.
Verde data loader:
Had some questions about data fields in Verde that don't exist in Voyager
Removed some unnecessary data from the Voyager extract, including PO number and fund codes (later wished had kept these two).
Encountered some problems, including blank fields for subscription histories (some caused by local record-keeping practices) - were able to make up with historical subscription info from EBSCO and Swets.
The data extract began with 46 columns. Anita reduced to 14, but actually used only 7 in load to Verde.
More reasons for the decision to transfer Voyager acq data to Verde:
--Easier for non-experts in Voyager acq client to search and find needed information
--Better type categories for journals, 3 types of database, ebooks, locally defined materials; different kinds of subscriptions - electronic, e+print, locally-defined fields
--Automated staff notifications
Mapping Voyager to Verde:
Lookup fields: ISSN, title
Straight match: Vendor, PO number to line item ID, price, subscription type
No direct match: subscription dates, subscription types
Data loader vs manual entry:
Titles didn't always match - journal titles were often okay, but packages rarely had the same name, and some titles had missing ISSNs - so how long would it take to normalize the data?
Also behind with Verde data updates such as moved packages (Haworth -> Informaworld, etc)
So determined that it was quicker to enter data by hand - staff could enter while doing other updates and creating records.
Developed procedure for data entry - rearranged Excel sheet to match Verde process (oldest first)
Began entering in Jan 2010 for testing; once procedure was established, other staff began entering in Feb 2010; they are still entering but are about 100 titles from being done.
Some problems:
Interface changes (Haworth > Informaworld), incomplete info from Voyager on migration from e+print to e only, 2010 subscription info.
Positives about entering data manually:
Quick data entry once practiced; opportunity to clean up Verde data; opportunity to determine new ways of recording data
Things they'd change:
Get a time machine: historical decisions for record keeping in Voyager had a big effect on what data could get for Verde and how could get it
Involve more people for data entry as soon as procedure is tested
Ideas and recommendations:
--Identify how much you'll want to load - amount of history is a moving target depending on how long it takes to get the data in
--Format and clean up your Voyager data now to get around some of those historical decisions - example - distinguishing print subscriptions from electronic ones
--Tag manually-entered and EDI data within PO Notes and Line Item Notes fields - example - for subscription start and end dates
Q&A:
--Currently using Verde as primary acq tool for eresources so can use workflows, then put into Voyager later
--Voyager line item ID number should be loadable into Verde with the data loader - needs a good match point but is a good way to start small
--Danger of creating duplicate acquisition records in Verde using the acq loader - if you load data once, then do it again, Verde assumes you want a new record.
Comments