Match, merge, overlay

Matching, merging, overlay

National encoding level	Institution Priority
special topic: Master record and holdings display when records are suppressed

When records come from local sites to Summit (“central”), the system assesses a number of variables to decide whether to update some bibliographic or institution information in the union database bib. record, add the record, merge incoming institution information (holdings) to an existing record, replace an existing bib. record with a new one and add all institution records to it, or split a record (used for deletes, attaching institution records to other bib. records, getting another bib. record from a local site and adding it to Summit).

These are some of the variables that the software considers in matching, merging, and splitting:

001 field: Is the 001 field a legitimate OCLC number?
Does the record already exist on central?
- 001field matches, either legitimate OCLC number or made-up local record no.
- ISBN matches
Origin of transaction
Is the sending site the owner of the bib. record ("LOC" in bib. fixed field; or see 945, subfield o=1) ?
Nature of transaction
- Is this coming as a delete?
- Is this coming with change in BCode3?
  - suppress code present
  - no suppress code (is it to un-suppress?)
- Institutions attached: Are there others institutions attached to this bib. record? (other 945 fields)

A record may remain the master record even when the owning site codes it for suppression, or has no holdings that display (local bib. record may have no attached records, or holdings may be suppressed). For more detail see Record contribution and display.

Matching

OCLC #. When a bibliographic record comes to central, the 001 field is compared to entries in the OCLC record # index. When the system identifies a match -- that is, the 001 field in the incoming bib. record matches the 001 in an existing record -- the system will overlay, or merge, the records.

ISBN #. When a matching OCLC record number isn't found, then the system

compares the ISBN [1](020 field) of the incoming record to numbers indexed in the "i" index. [If no matching ISBN is found, the record is added, or inserted.] If the system identifies a matching ISBN, it
looks at the OCLC number fields in both records. If the records have different OCLC numbers, the record is not considered a match and the record is added. If an ISBN match is found and at least one of the records does not have an OCLC number, the system then
compares the first three significant words of the title.

So, if (1) the ISBN matches, (2) at least one record doesn't have an OCLC number, and (3) the titles match, the system has made a match, continues with the overlay, or merging, process.

Merging

Overlay criteria. To select the single record for the union catalog, the Innovative software looks at Encoding level and a Priority List of Libraries (see next page).

"First in". When the incoming bib. record has the same encoding level and institution priority codes as the record already in the union catalog, the institution record for the incoming bib. is attach ed to the record already in the database.

1. ENCODING LEVEL (ENC LVL). Central uses Innovative's suggested Encoding level rule list. How it works - example: Record A has enc lvl code "blank", ranked as Priority 10 , top priority; Record B has enc lvl code 8 for CIP, ranked as Priority 5. Record A is used for the union catalog, and institution records attached to Record B are attached to A.

Encoding level	Meaning	Priority group
[blank]	Full level	10 (highest)
I 4	Full level, input by an OCLC library Core level (effective with R2001)	9
1	Full level, material not examines	8
L	Non-LC & non-NLM loaded from tape	7
K J 2 M	Less than full level, input by OCLC library Record deleted by LC from MARC file Less than full level Less than full level, tape-loaded	6
8	CIP prepublication data	5
5 7	Partial level Minimal level	4
E W 3	System-identified error in LC MARC record Warning, possible error in LC MARC Abbreviated level (effective with R2001)	3
U Z ???	Unknown Not applicable [any code not listed, such as erroneous or new codes]	2

2. INSTITUTION. When two records have the same Encoding Level code, one record is selected based on Institution Priority. Orbis members adopted a priority grouping based on overall record quality, which includes these concepts:

·Completeness: Retention of all fields vs. systematically (through profiling) or selectively (individual editing) removing fields; maximum access to bibliographic information available through all indexable fields

·Accuracy: Initial and ongoing authority work: author, series, subject headings; CIP review and completion; compliance with AACR2, LC, OCLC, and other national cataloging standards for content, field and subfield codes, and indicators.

10 (high)	UO, OSU & COCC, UW
9	PSU
7	EOU; WOU; Linf; SOU; GFU; Reed; UPS
6	SPU, MHCC
5	WU; OIT; UP; L&C, Marylhurst, WWC
4	Pacific, CWU, WWU
3	Whitmn, OHSU, WSU, TESC & SMC, EWU, UW-Law, PCC, Clark
2	CRL, LCC

[1] Only the first 10 significant digits of the ISBN. Hyphens and all data following the first non-numeric character in the ISBN (if any) are stripped prior to matching.

updated: November 15, 2007