Database Preparation Services

Item Field Build & Barcode Label Production

Item Field Build

Circulation control systems use "item records" to identify library materials. Once the database is loaded, item records are maintained by library staff using the local system. During the initial database load, item records are created by the local system from "item field" information. While the library's database vendor can build "item fields," it is the local system vendor that converts item field data into the actual item record used by the local system.

Different local system vendors use different variable fields to store item field information—e.g., Geac uses the 049 field, Sirsi the 949 field, and Innovative Interfaces a 940/945 field combination. Typically, an item field will contain some or all of the following information: 1) item number (usually 14 digits); 2) call number; 3) copy and volume information; 4) holding library, branch, location, or collection code; 5) media code (e.g., book, audio cassette); 6) statistical category code—used to generate reports of materials circulated within a category or classification range (e.g., detective stories, sociology books); 7) fine code; and, 8) period of circulation code.

Because each local system vendor has its own special requirements, the item field should be constructed using a format prescribed by the local system. To build item fields the database vendor requires the specification used by the local system, as well as information about the library's cataloging practices and profiling. Data tables are prepared that map the contents of each subfield in the item field.

The complexity of an item field build can be appreciated by examining a single item field data element, the call number. It would seem a trivial task to move a call number from the catalog record into the appropriate subfield of the item field, but the operation is more involved than it first appears.

To format an item's call number correctly, the database vendor must take into account bibliographic and holdings data found throughout the catalog record. For example, it is a standard practice of OCLC participants to use holding library codes to generate automatic stamps above or below call numbers. Thus, the four-character holding library code ZZZR may be profiled to print the automatic stamp REF above the call number. Similarly, the library may not use the same classification in all its holding libraries or collections. The call number field selected from the catalog record may depend on the holding library code in the 049 field. To illustrate, holding library code ZZZJ may be profiled to print the call number in the 092 field while holding library ZZZA may be profiled for use with the call number in the 090 or 050 field.

Some libraries may designate the location of an item by inserting a "stamp" prior to the call number. For example, prior to the class portion of a Dewey call number, the library may have inserted the word "Ref." It is necessary to remove this stamp from the call number given in the formatted item field because leaving it there will destroy the indexing of that call number within the local system's call number index. In addition, the presence of this stamp ("Ref.") may be the only indication in the record that the location code in the item field should be set to reference as opposed to the main or default location code.

Even the size of a book may affect its call number. Many libraries shelve oversize materials separate from regular size books. Oversize materials often have an automatic stamp (e.g., OVERSIZE) placed above the call number. This stamp is generated whenever the book's height exceeds a predetermined number found in subfield $c of the physical description field.

For OCLC libraries, the database vendor also needs to know if the "suppress x" option has been used to produce catalog cards for titles that are not classified. Other considerations, including how "input stamps" should be treated, which call number to use when there is more than one with the same tag (e.g., multiple 090 fields in the record), and call number selection hierarchy (e.g., 099/090/050 for LC; 099/092/082 for DDC), must be rigorously defined if the call number on the piece is to match the call number displayed to the user within the local system and the call number printed on the smart barcode label.

Of course not all database vendors bother with the above details or even attempt to build library and local system specific item fields. Some create a single generic item field for each catalog record, regardless of the library's holdings data or the local system in which the item fields will be loaded.

The "one-size-fits-all" approach to item field creation simplifies the database vendor's work by transferring the difficult part of the job back to the library. Failure on the part of the vendor to create comprehensive item fields based on the library's holdings information can undermine the success of the library's automation project. After the library's database is loaded, library staff will have to manually create item records for the ones not built by the vendor. Moreover, because smart barcodes are a byproduct of the item fields, staff must use dumb barcodes for copies and volumes ignored by the database vendor.

Many of the uncertainties and unanticipated problems associated with library automation can be eliminated by selecting a database vendor who recognizes differences among library databases and accommodates the special requirements of the customer. At LTI, each item field build is a custom job that takes into account the bibliographic and holdings data actually present in the record as well as the item field specification used by the local system. Parsing of detailed copy and volume holdings data from subfields $c, $p, $v, and $y in OCLC 049 fields is standard procedure. If holdings information indicates there are five volumes associated with a title, five item fields will be created and five smart barcode labels will be printed.

Gap Tape Processing

Typically, database preparation services are needed at two points in the automation process. First, the initial or main database of records receive deduping, item field build, smart barcode labels, and authority control. About three to six months later the library will require the same services for a much smaller number of records covering the period between the initial load cutoff date and the date the library begins to download records directly into its local system. These records are commonly referred to as "gap" or "catch-up" tape records.

Canceled or withdrawn items and/or bibliographic records appearing on the gap tape must be deleted manually from the local system by library staff. In addition to repeating the same processing procedures used in the main database, for OCLC libraries the database vendor must provide the library with a printout of Cancel transaction records. Working from this printout, library staff will first delete any item records attached to the bibliographic records, then delete the bibliographic records. It follows that a library would not want to initiate or continue a massive cancelling or withdraw project during the gap tape period.

OCLC "gap" tape records should be processed using the same procedures as used in the main database. This is not a good time to initiate new procedures. Hold off altering processing procedures until after the new local system is fully operational and library staff have received cataloging training on the new local system.

0ne exception to the above advice is entering holdings information during the gap tape period. If the library follows the practice of inputting complete holdings data for every use of a record, then duplicate item fields and wasted smart barcodes will be produced from processing gap tape records. To avoid this the library should input only new holdings data during the gap tape period.

Barcode Production

Automated circulation control systems use barcodes (a series of printed lines and spaces of varying widths which represent numbers, symbols, and/or letters) to track items through the circulation process. While some systems can accept either barcodes or optical character recognition (OCR) labels, barcodes are easier to print and scan.

Barcoding is an expensive and labor-intensive activity. It is also the first critical project library staff undertake prior to automation. While not a glamorous undertaking, a well-planned barcoding project is an important step in helping to establish favorable staff and user attitudes toward the new system. Procedures concerning the recording and resolution of barcoding problems should be documented in advance of the barcoding.

Barcode Symbology

Many different barcode schemes are used in business and industry. The Universal Product Code (UPC), now ubiquitous in supermarkets as a means to control inventory and pricing, is perhaps the best known barcode symbology.

Only two barcode symbologies are commonly used in libraries: Codabar (or Code-A-Bar) and Code 39 (or Code 3 of 9). In library applications, Codabar generally consists of a string of 14 digits. The first digit represents a patron or item. That is, a first digit of '2' tells the circulation software that a patron number is being scanned; the number '3' signifies that an item is being scanned. The next four digits (digits 2 through 5) are the institutional identifier. The local system vendor either assigns this number or the library can make up its own institution code. For example, the library might use the last four digits of its telephone number or the last four digits in its extended nine-digit ZIP code.

The next eight digits (digits 6 through 13) define the sequence ID number. Most smart barcode sequences begin with the number 1—i.e., 00000001. The final digit (digit 14) is a check digit used to verify that the barcode number has been scanned accurately. An eye-readable form of the item number is usually printed beneath the scannable number. To improve human readability, spaces are inserted between digits 1 and 2, 5 and 6, and 10 and 11.

Code 39 may contain fewer than 14 characters. For example in a ten digit number, the first digit indicates whether the code represents a patron or an item, the next 2 digits define the institutional identifier, and the last 7 digits identify the applicable code. While in theory Code 39 applications do not require a check digit, in practice many do add a check digit. Alpha characters can be used with Code 39 but they should be avoided if possible.

Both Codabar and Code 39 are self-clocking and permit bi-directional scanning. Library materials should not be barcoded before the library has chosen its local system.

Smart vs. Dumb Barcodes

Barcoding is the process of assigning a unique item number to each piece that can circulate and linking that item number to an item record, which in turn is linked to a bibliographic record. Library barcodes labels are classified as either "smart" or "dumb." Dumb barcode labels are also referred to as "generic" labels. Smart barcodes are always applied prior to bringing up circulation control, making it possible for library staff to concentrate on mastering the new system rather than the arduous task of making sure the right barcode gets put on the right item. Dumb barcodes can be applied prior to bringing up the system or at the point of check-out.

Smart barcodes have item numbers that are assigned by computer during the item field build on the basis of copy and volume holdings appearing in the bibliographic record. Because they are preassigned by machine to specific items in the database, their use avoids the tedious work of manually linking barcode numbers to item records. Smart barcodes come in a variety of sizes and styles and cost about four cents per label. Since proper shelf arrangement is critical to efficient smart barcode application, a preliminary shelfreading project should be considered prior to applying labels.

Smart barcodes entail extra costs in both database processing and label production and their use can reveal problems that will later demand attention. Smart barcodes have the added advantage of providing a quasi-inventory of the collection during their application. Whereas smart barcodes alert the library to database problems early on, dumb barcodes do not reveal these problems until the linking step occurs. From that point onward the clean up and correction process are similar.

Dumb barcodes are less expensive to produce and can be applied at random throughout the collection. However, depending on the nature of the collection, they may be more costly when one factors in the time necessary to link the barcode numbers to item records. The great advantage of dumb barcodes is that they can be applied by unskilled workers prior to circulation or even at the point of circulation. Their disadvantage is that each barcode has to be manually linked to its item record.

Even if the library has opted for smart barcoding, it still needs to acquire dumb barcodes for new titles and for those items for which a smart barcode label has not been produced. The latter can be caused by either the database vendor ignoring the library's actual holdings or the lack of holdings information in the catalog record. Libraries should not order more than a three-year supply of dumb barcode labels.

Opinion varies as to whether smart barcoding is always the best approach to linking library materials to item records. As a rule, smart barcoding is recommended for collections having specific call numbers and only a single copy of most titles. Academic and special libraries will almost always benefit from smart barcoding. Public libraries, having multiple copies and non-specific call numbers (e.g., FIC), may encounter problems in applying smart barcodes, particularly if the bibliographic records do not contain adequate holdings data. Dumb labels are recommended for very small databases, e.g., under 10,000 records.

Barcode Location

The location of barcodes on library materials needs to be determined in advance. Given the expense of barcode labels, most libraries will only want to purchase a single label for each item. There are advantages and disadvantages to where the label is placed on the item. Affixing the label to the inside back cover endpaper offers good protection and, if book pockets or date due slips are used, increases check-out efficiency. Label suppliers recommend that labels be placed on the inside rather than outside cover or spine of books.

Placement of the label on the outside of the cover facilitates inventory and in-house scanning without having to open the book or in some cases even remove it from the shelf. As with other aspects of the automation process, consult with your local system vendor and other experienced users of the system before making barcoding decisions. Large libraries might want to review an Association of Research Libraries publication, Spec Kit 124 (May 1986) titled "Barcoding of Collections in ARL Libraries."

Pitfalls to Avoid

Opinion also differs on the wisdom of assigning dumb barcodes "on the fly" as part of the cataloging operation before the local system is selected. This is accomplished by applying a dumb barcode to the item as it is being cataloged. The barcode number is either keyed into the holdings field of the catalog record or "wanded" in with a barcode reader. Those recommending this practice argue that the library is getting a head start by reducing the number of barcodes that need to be applied later.

LTI's experience has been that this practice always makes more work for library staff and limits the institution's flexibility. Whatever benefit results from having barcoded a certain portion of the collection in advance is more than offset by the new problems it creates. Since the library does not have online access to its database, there is no easy way to maintain barcode numbers assigned to records. In addition to the extra work of getting the barcode number into the record at the time of cataloging, staff will need to keep a log of replacement item numbers for labels removed or destroyed by patrons. When a record is reused, staff may need to rekey the barcode number(s) into the latest use. Database vendor programming costs will be higher because two different methods of building item fields are required. Moreover, the library runs the risk of adopting either a barcode symbology or check digit that will not be supported by the local system vendor.

Your library may find it expedient to order dumb barcode labels before the database vendor has completed the item field build and produced the smart barcode labels. If so, a comfortable margin should exist between the last possible machine-assigned item number and the first dumb barcode item number. As a guideline, academic libraries can estimate the number of item fields by multiplying the number of records remaining after deduping by 1.1 or 1.2. For example, if the library has 100,000 catalog records following deduping, one might project 110,000 to 120,000 item fields. Because of multiple copies, public libraries can have a higher ratio of item fields to records, in some cases as high as 1.5.

Before ordering dumb barcodes the library must confirm with the database vendor what the start sequence number should be for the dumb labels. Using the example above of 100,000 catalog records and an anticipated 110,000 to 120,000 item fields, the library might want to start the dumb barcode sequence at 300,000.

Thus, if the library is using a 14-digit item number and its institution code is 4921 and the first machine-assigned sequence number is 1, the first item number will be 3 4921 00000 001C, where the letter C at the end is a check digit whose value is a number between 0 and 9 calculated from the values of the 13 numbers preceding it. To eliminate any possible overlapping item numbers, the first item number assigned to the dumb barcodes might be 3 4921 00300 000C. The barcode producer must also be told how the check digit is to be calculated, e.g., MOD 10, MOD 43.

Barcode Label "Driver" Tape

Smart barcodes are printed using a photocomposition process from data written to a barcode label "driver" file. Data used to format the labels comes from the item field and other fields in the library's catalog record. Because much of the information used in the label is derived from the item field, smart barcodes cannot be produced without first creating item fields. If the library has not yet selected its local system, the database vendor cannot format the item fields and, without item fields, smart barcodes cannot be printed. An exception exists for those database vendors that create generic item fields without regard to library holdings or the local system item field specifications.

In addition to data taken from the item field (e.g., the item number, call number, branch/location code), smart barcodes must contain sufficient bibliographic information to match each label to the piece. On a standard 2" (wide) x 3/4" (high) smart barcode label, two lines are available to print call number, holding library or location, and abbreviated author/title information. Other lines are reserved for the library's name, along with scannable and eye-readable versions of the item number. The only time the call number and related bibliographic information on the smart barcode label is useful is when the label is being applied to the item. Dumb barcodes are commonly printed on 2" x 5/8" stock and include the name of the library at the top, followed by scannable and eye-readable versions of the barcode number.

Prior to creating the barcode label driver tape, the database vendor must know if there are bibliographic formats (e.g., serials), types of materials (e.g., periodicals), or special collections (e.g., reference, rare books) for which barcode labels should not be printed. Barcode labels are usually arranged by holding library or collection and sorted by call number. Efficient smart barcoding depends on how closely the sorted barcode labels follow the shelving arrangement of the library materials.

Next: Additional Services