Helen Williams Assistant Librarian, Cataloguing

A.Rose, by any other name:

A workshop on name authority control

CILIP

Oct 2009

Helen WilliamsAssistant Librarian, Cataloguing

Retrospective authority control

(and the impact on current

authority control procedures)

Background

• LSE library founded 1896• First subject catalogues appeared 1931-32• Library houses over 4 million printed volumes (including journals, historical

pamphlets, archive materials).

• Authority control procedures had been varied over the years

• Result = library catalogue with a number of variant headings

Initial considerations

• Authority control group convened in 2006• Assessed user needs• Created authority control guidelines for staff covering

names, corporate names, subjects and series • Analysed current position• And possible options for improvement

Detailed considerations • In-house • Estimated at 21,000 person hours based on an average of 2

minutes to check each heading – equivalent of one person working on the project for 12 years!

• Outsourcing • Export entire catalogue to external company to do one off check

of names, corporate names, subjects and series. Headings are checked, validated against LC, alternations made, file returned and loaded back.

• Recommend regular checks of new records after project and accompanying error reports to assess future possibilities for in-house work.

Tendering process (1)

• Outlined background and scope

• Requirements for project

• Responses required from tenderers

Tendering process (2)

• Tenders were received from 3 companies

• Considered by working group

• Project awarded to Marcive

Preparation for exporting file to Marcive

• Detailed specifications completed in-house to send to Marcive

• Liaison with Marcive

• In-house preparation and planning

Exporting and test phase

• May 2007 – exported approximately one million bibliographic records to Marcive

• 2 weeks later received test file of 10,000 records as a sample for checking.

• Checked for corrected indicators/punctuation, 1XX/7XX, 440/8XX, 6XX fields changed to correct authorities, 245 GMDs correctly replaced, abbreviations corrected, and that no general corruption of data had taken place

Queries arising from sample checking

We checked the test sample so thoroughly that we came up with a number of queries to be submitted to Marcive.

– Basic spellings not corrected – eg Germay to Germany– Names not recognised for correction where no space has been entered

before the brackets eg Marks & Spencer (Firm) changed to Marks & Spencer Plc, but Marks & Spencer(Firm) not changed.

– Department not changed to Dept. on names where that would then correct the authority. Eg Department of Employment to Dept. of Employment.

– Headings which needed a qualifier added and it was not eg Fabian Society needed (Great Britain) to validate

– Incorrect authority attribution eg Daunton M. J. (Martin James), 1949- was changed to Daunton, M.J. (Martin J.), 1949- even though the LC authority record does not include 1949.

Marcive response

– Excellent response in terms of customer service

Quick and detailed

– But it highlighted that we had higher expectations of the automated process than was actually achievable

– How we dealt with this

Return of cleaned data

– A few weeks after checking the data and providing Marcive with written confirmation to go ahead we received 21 bibliographic files which comprised our entire catalogue and 10 files of authority records.

– IT dept had intended to load all files to the test server first, but took 7.5-9 hours per file and was therefore impractical.

– Decided to check 2 bib files and 1 authority file on test server … but …!

Problems

– These projects are never as straight forward as one hopes!

– The test server indexes suffered under the strain of so much data and we had to wait for Ex Libris to carry out a regeneration of the indexes.

– Then we had a problem with Russian diactritics supplied by Marcive not being accepted by the character set validation on Voyager.

– Followed by a problem re-loading and re-indexing data in the live server. Even changing our in-house protocols the process still took 143 hours!

Ongoing services

– January 2008 and time to think about the ongoing services Marcive were providing for us.

– Overnight authorities service and• Multimatches report• Unrecognised main headings report

– Notification service • Global headings change

What is Global Headings Change?

– When a change is made to a 1XX field in an authority record then the change is sent to the GHC queue so that by using a particular command you can set off a database clean up so that all related bib records are also corrected.

– Overnight authorities service and• Multimatches report• Unrecognised main headings report

– Notification service • Global headings change

What is Global Headings Change?

– When a change is made to a 1XX field in an authority record then the change is sent to the GHC queue so that by using a particular command you can set off a database clean up so that all related bib records are also corrected.

– Overnight authorities service and• Multimatches report• Unrecognised main headings report

– Notification service • Global headings change

In-house considerations

• With these ongoing service we considered whether we no longer needed our stringent in-house authority validation checks at the cataloguing stage.

• On reflection we were confident that existing authority control guidelines needed to continue

Follow-up to initial project

• In addition to embedding the ongoing services into our in-house work we needed to do some work to tidy up the headings Marcive had been unable to change.

• Scoped out work and employed a temp for 22 weeks working 4 days a week.

Project work (1)

• Multiple reports returned from Marcive so had to establish order of priority which we based on user satisfaction levels.

• High – unidentified subject headings and unrecognised geographic main headings, followed by in-house report of wrong or missing indicators

Project work (2)

• Medium priority – Personal names, corporate names, multimatches, geographic subdivisions

• Low priority – Uniform titles, series headings, meeting names.

Work completed

• Our temp spent around 630 hours working on the project and completed 37,514 records

• Approx one record per minute

• Estimate an average of 2 headings needed changing per record, so approx 75,000 headings were corrected.

• .

Work remaining

• Remaining personal names• Corporate names• Multimatches (all fields)• Remaining geographic subdivisions• Series names• Meeting names

But – more complex to deal with and would show lesser return for the time and cost involved.

Pros and cons of more in-house work

• Advantages to completing the work = accurate catalogue, minimal errors or confusing entries

• But, in reality, impossible for an institution with such an old catalogue to be completely free of errors.

• Sample testing shows that authority records are not available for over 90% of the remaining headings.

• Total time taken to complete medium priority reports alone would

be 15,437 hours! • Remaining headings are not considered high priority to aid user

retrieval

Thanks…

• Due to those in the Bib Services team who took part in testing data

• The senior library assistant for overseeing the temp

• The support of our IT dept – our IT contact has a cataloguing background so had an invaluable understanding of both the technical and practical aspects of the project.

The results • Our catalogue is more consistent and has less errors

• In-house services and ongoing Marcive services ensure that the benefits of this work are maintained

• The project has increased the profile of authority control in Bib Services

• Library catalogue continues to get a high score on student satisfaction survey