In Library And Information Science

Getting Metadata Right Is a Full-Time Job Nobody Warns You About

I still remember the first time I cataloged a monograph series that turned out to be three separate works bound together under one publisher's mistake. The MARC record I built took me four hours because the source description was literally contradictory on every level. That kind of problem shows up more often than you would think. The field sounds simpler than it actually is. People hear "library" and think shelving books. The real work sits in cataloging, classification, metadata management, and information architecture. Those are the gears that make everything else function. The backbone of most library work is MARC21. It has been the standard in North American libraries since the early 1970s, when the Library of Congress converted its card catalog to machine-readable format. If you are building records from scratch, you will work with fields like 245 for title, 260 or 300 for publication and physical description, and the various subject heading fields like 650 and 651. The exact configuration changes depending on your catalog system. Alma, Sierra, Koha — each handles subfield delimiters and indicator logic differently. You learn it through repetition and error.

Classification is another layer that gets underestimated. Dewey Decimal Classification collapses a lot. For academic and larger public libraries, Library of Congress Classification is the default. But LC is not free. You need access to the LC web resources and, increasingly, the online classifications. The call number construction alone can take a trained librarian ten to fifteen minutes per item when the subject is borderline between two categories. Subject heading control depends heavily on Library of Congress Subject Headings or Sears List of Subject Headings for smaller collections. RVM is standard in Quebec. The British Subject Headcomes into play in UK institutions. Getting these wrong means your users cannot find what they are looking for, and fixing bad subject entries retroactively is genuinely painful.

A Problem I Actually Encountered

Last year I worked on a digitization project where we received box after box of uncataloged institutional reports. Roughly two hundred items. No ISBNs. No ISSN. Very inconsistent title pages. Some were produced by government agencies, some by nonprofit organizations with different name changes over the decades. The biggest headache was corporate authorship. One organization had merged with another in 2018, and the reports spanned thirty years. I spent a full day just cross-referencing authority files to get the name heading right across all variations. The workaround was building a quick lookup table in Excel mapping each variant form to the authorized heading, then running a batch macro in SirsiDynix Symphony to update the records. Saved me probably a dozen hours of manual editing. It is not the most elegant solution, but it works for medium-sized backlogs when you do not have budget for external cataloging services.

Get the Full Details

Applied Research in Library and Information Science - kaufmanpress

What Nobody Tells You About the Work

One counter-intuitive thing about this field is that speed and accuracy are often in direct tension. Rushing a MARC record to meet processing targets results in records that break discovery. A missed 650 field means the item simply does not show up in subject searches. Users will blame the catalog. They will not blame the cataloger who rushed it. Another thing: RDA is the current descriptive standard, but it is not universally adopted. Many libraries still operate under AACR2 rules for legacy reasons. Your choice of standard affects every single field you fill in. Mixing them in a single record causes parsing errors in OAI-PMH harvests and broken exports to partner systems. Always check what your consortium agreement requires before you start a batch project. Koha and other open-source ILS platforms are getting better, but they still struggle with certain advanced MARC features like nested subfields in field 440. If your collection has a lot of series material, plan around that limitation. I learned this the hard way when a colleague configured a Koha instance and then discovered that series statements from fifty thousand records were rendering incorrectly in the OPAC.

Downsides and When This Approach Fails

Manual cataloging is slow. Even an experienced cataloger typically processes between three and eight complete records per hour depending on complexity. For large acquisitions, that is a bottleneck. Library suppliers like Baker & Taylor or Brodart offer cataloging services, but they vary in quality and their records may not match your local policies exactly. OCLC QuickMarc helps when you have matchable records in WorldCat. It can cut processing time dramatically — sometimes down to minutes per item. But it fails when dealing with non-English materials, self-published works, or items that simply do not exist in WorldCat yet. You still need human judgment for those. Metadata quality is also a systemic problem across digital repositories. Dublin Core is widely used because it is simple, but its simplicity becomes a liability when you need precise date values, place names, or controlled vocabulary. Researchers who depend on your collections will notice the gaps. They just will not tell you.

Practical Steps for Getting Started

If you are new to this, start with a solid understanding of MARC21 structure before you touch any catalog system. The Library of Congress offers free training at loc.gov/marc. Work through the basic exercises. Then pick a small collection — twenty to thirty items — and build complete records from scratch. Compare your work against existing OCLC records. The differences will teach you more than any textbook. Learn your classification system's scope notes. They exist for a reason. Skipping them is the fastest way to put a book in the wrong call number slot and confuse every user who searches by subject. Build relationships with your technical services colleagues. Cataloging is not a solo sport. Disagreements about subject heading choices, classification boundaries, and naming authorities are normal. A second pair of eyes catches mistakes that you will not see because you have been looking at the same record for too long.