McDanold, "The Differences Problem: Or why Consistency in Metadata is Critical in the Discovery Process"
Download as PPTX, PDF0 likes366 views
The document discusses the importance of consistency in metadata for effective ebook discovery across various platforms and vendors. It highlights the challenges faced with traditional and modern searching methods due to differences in metadata presentation and the impact on user searches. Recommendations for improvement include adopting controlled vocabularies, following best practices, and ensuring thorough communication among vendors.
McDanold, "The Differences Problem: Or why Consistency in Metadata is Critical in the Discovery Process"
- 1. The Differences Problem Or why consistency in metadata is critical in the discovery process Shana L. McDanold
- 2. First A few caveats… 2
- 3. Inthenotso distantpast… There were two main options when searching for ebooks: 1. Search each individual vendor’s website/database 2. Load MARC records (one record for each title) into the catalog for each vendor 3
- 4. Inthenotso distantpast… Problems with this approach: Loading records is a LOT of work and requires regular maintenance Massaging/editing/enhancing metadata; loading; updates; replacements; deletes Number of records/titles to load Lack of records available for loading Records come from numerous places and each vendor requires a different procedure to download files Tracking titles in multiple places (duplicate work) 4
- 5. Now:more options… 1. Search each individual vendor’s website/database 2. Load MARC records (one record for each title) into the catalog for each vendor 3. Integration of various vendors metadata into discovery layers via APIs and linked data rather than importing records into the catalog 4. Federated search tools that index multiple databases (e.g. unified index search tools) …but are more options better? 5
- 6. Thegoodand thebad GOOD: fewer places to search (possibly even only one) most public libraries, while they have other ebook databases, will have a single integrated discovery layer BAD: MORE places to search BUT discovery is still a challenge no matter which search option you choose, and those challenges are centered around: METADATA 6
- 7. Printbook 7
- 8. Ebook 8
- 9. Differences? ISBN Subjects Title Author Date 9
- 10. Printbook 10
- 11. Ebook 11
- 12. Differences? ISBN Subjects Title Author Date 12
- 13. Printbook 13
- 14. Ebook 14
- 15. Differences? ISBN Subjects Title Author Date 15
- 16. Differences defined Differences in description Current vs past rules and guidelines; RDA provider neutral vs individual vendor records Differences between vendors for same title Differences in how data is entered/presented Record proliferation Related to metadata differences: records cannot be “collapsed” because the discovery layer doesn’t recognize them as the same Different vocabularies and identity databases 16
- 17. More differences Missing metadata/missing records Data changes/updates Branding or custom text/collections 17
- 18. Whydothese differences matter? How people search Keyword - forces dependency on keyword indexes Follow links - if you click on the subject search for Obama, Michelle, search results include only print books (no ebooks) Limits/facets - dependent on metadata, both visible and invisible (coded) Missing metadata Discovery layer exposes ALL the metadata (good, bad, missing) All means items get “hidden” because they’re not findable. 18
- 19. How dowefix it? CONSISTENCY use of controlled vocabularies and existing authority databases (name matching, subjects, etc.) Use existing metadata sources Follow standards and recommended/best practices Communication Data points complete consistency across vendors 19
- 20. Questions? 20
Editor's Notes
- #3: Usually differences are a GOOD thing, providing diversity; but not in this case Caveat: speaking from a public library perspective mainly; although most of the issues public libraries have are present in academic environments; differences are resource types and focus on currency/popularity of materials (collection is more ephemeral than permanent) BUT my background is serials and nonprint format cataloging – been dealing with managing metadata/cataloging for ejournals and ebooks for almost 2 decades now My philosophy: job of cataloging/metadata is to make stuff findable, which includes unique identification of resources I don’t believe in the “perfect” record If it’s not wrong, leave it alone (don’t delete data, just exclude it from indexes…you may want it in the future) When editing: Fix errors or delete if wrong Add access points Enhance content/description (add value) Make it pretty
- #5: Number of vendors increased – more complex more time Each vendor: different procedure for downloading; different edits (some need proxy added, some don’t); files may be in various formats and require conversion to MARC Tools to help streamline (MarcEdit – TASK LISTS saving the edits for each vendor are a savior) BUT still very time consuming Multiple places: ERM and the Catalog and possibly the vendor website – have to keep in sync
- #7: Looking at a single search option for ebooks and print books, where an API is used to search both ebook vendor and the catalog in one search So lets look at examples – examples are current popular titles or authors
- #8: Who’s watching the show on Netflix?
- #10: ISBN: this is often a key match point for OpenURL resolvers or other API/linked data tools Title: ebook version is incomplete Author: translator is missing, an issue when looking for a specific translation or if searching by translator name Date format – indexing issue – how does your system handle dates?
- #13: ISBN: this is often a key match point for OpenURL resolvers or other API/linked data tools Title: ebook version is incomplete Author: indexing issues; identity management/authority control issues Date format – indexing issue – how does your system handle dates?
- #16: ISBN: this is often a key match point for OpenURL resolvers or other API/linked data tools Subject: where’s DC?? Title: ebook version is different Author: indexing issues; identity management/authority control issues Date format – indexing issue – how does your system handle dates? Do you see a trend yet?
- #17: Description: AACR2 vs RDA – fundamental change in how you approach describing a resource Provider neutral – one records for ALL online versions of a title (formats, platform, etc.) – just have multiple links/URLs to various options; Hard to do that with APIs/linked data tools Date format, author format (last, first or first last?) Proliferation: more vendors = more records We get patron complaints about ebook display all the time Different vocabularies and identity databases – name formats, subjects, locations, etc. Creates indexing and filing issues; split indexes
- #18: Missing: sometimes records just don’t appear – API/linked data tool errors, delays, Data changes: records get “out of sync” – print book may be complete but ebook is still minimal/prepublication Branding: can’t add custom text to create collections, or other data to ebook records; limits to control over display and what data is included – stuck with what the vendor sends/makes available
- #19: Forcing dependency on keyword indexing or indexing of the WHOLE records – specific author indexes, etc. become not useful How people search: Subjects/identities – FORM matters “see also” Collections Links – find something the want/like, follow links to “similar” or “like” items using subjects, authors, etc. (internet rabbit hole…) Limits/facets – such as format, publication date, location, etc. Missing metadata – subjects, ISBN, names, locations, etc.; lose match points; may result in records not appearing – search ISBN and the ebooks don’t show up Discovery layers – good at exposing EVERYTHING (great way to identify database cleanup projects…)
- #20: Communication – between libraries and vendors Data points – more is better, even if they don’t display