J. Abernethy
Tracking in Widener Library
One primary focus of creating a more efficient Widener Library involved being able to detect book movement within the large, twelve story building. In such an immense environment, it is inevitable that books would be lost due to the misshelving, mishandling, and misplacing of various items. In fact, we calculated that tens of thousands of books are misplaced each year, with no guarantee of returning to the system in a logical manner. On many occasions, patrons would not be able to find a book of interest, due to itís being placed in an incorrect location.
Tracking through Widener Library was roughly divided into four sections: 1) a system which would rely on multiple scan checkpoints similar to Fed Ex. 2) A logical and reasonable method of assigning books to their correct locations (e.g., shelves, carts, studies). 3) The ability to check the status of a book in order to determine whether it has been placed in the correct location. And 4) a new labeling and identification systems which would make the other sections more feasible to the library staff. My main focus for the Widener project concerned researching and identifying possible alternative identification and labeling systems which would improve and promote any new tracking system, as well as assist other project tasks such as reshelving and sorting.
Current Identification System
To begin our pursuit for an effective identification system, the current barcode system of Widener Library was examined and scrutinized. Widener incorporates a 14 digit barcode printed on a 2" X 5/8" label. The label is located on the upper right hand corner of the inside back cover of each book. Data Composition Inc., the labeling firm which Widener Library employs, uses code 39 as their barcode system of choice. Code 39 is a fairly standard alphanumeric symbology which uses 5 lines and 4 spaces to encode each character. When using a 7.5 mil dimension, 9.8 characters can be printed in 1 inch of space with code 39.
Widener Library has adopted the use of CCD, or Charge Coupled Device scanners. CCD scanners are fairly cheap in comparison to laser scanners, and are not composed of moving parts, which attributes to its longevity. One limitation is that the CCD scanner window must be longer than the barcode itself for a proper read.
Main Objectives of Our Identification System
Upon investigating different labeling and identification schemes, as well as the current system, the following objectives were considered:
More importantly, an ID system capable of maintaining an inventory at a smaller scale would be of great value to tracking the flow of books in the studies. The current study system lacks efficiency in that the professors or graduates students must fill out forms in order to properly charge a book. This becomes a time consuming process when checking out several books at one time. It should also be noted that the study owner may forget to fill out the correct forms for each book. This inevitably leads to books essentially lost to the studies, without any record of its removal from the Widener stacks. And since there is virtually no way to distinguish whether a book has been charged or discharged to a study, there is little chance for the library staff to detect these "problem" books.
However, if our ID system would allow the staff members to scan in books at a very rapid pace, then all the books in the studies could be read on at least a weekly basis.
"Problem" books would be detected by the weekly inventories of the studies, and either be properly charged to the study, or returned to the correct location in the stacks.
The following identification and labeling options were investigated and researched specifically for the Widener Library ideals and requirements listed above:
Redundant Barcodes on Outside of Book
The most obvious and logical labeling system is nothing more than a simple extension on the current Widener barcodes. Adding a redundant barcode to the outside of the book will surely take into account lost or damaged barcodes as there will be an identical code available for scanning. Furthermore, the barcodeís location on the outside cover would make scanning a physically easier process because opening the volume would no longer be necessary. It should also be noted that originally Widener Library employed the use of "smart barcodes" (barcodes which include book information such as title, call number, author, etc.) on the outside of the book during the original migration to labeling the volumes.
The exact location of the second identical barcode would be on the front cover, oriented vertically, and close to the spine. This position was decided upon two criteria: 1) When reshelving books, it is more customary to face the front of the book, thus making scanning easier, and 2) Library staff would be able to pull a volume partly while itís still on the shelf and scan the barcode. Therefore, a scan could take place without actually removing the book from the shelf.
Two major concerns need to be considered when employing redundant barcodes in this manner. First of all, the location of the barcode may conflict with the Harvard Depository barcode, which is mainly used to signify whether a book belongs in the Widener Stacks, or to the Depository. Fortunately, the new barcode would be easily discernible as it is oriented vertically as opposed to the HD barcodeís horizontal position, and the formatting for the two barcodes are easily distinguishable from each other. Secondly, the aesthetics of the books are a concern as many users, as well as the library staff may not appreciate having a black and white code on the front cover of every book. While it is true that any code placed on the outside of the book would detract from its appearance, the location of the barcode would not cover up any pertinent book information. Furthermore, since call number labels, as well as the HD barcode are currently placed on the books, it is likely to assume that an additional barcode would not tarnish the bookís appearance significantly.
Although placing barcodes on the outside of the book solves our major concerns for the new labeling system, it does not address the ability to inventory sections of the library in an efficient manner. It will undoubtedly take a great deal of time to scan in several volumes at once, because acquiring the book information obligates the staff member to partially pull each book so as to make the front barcode accessible. In our preliminary tests, this process was decidedly slow and tiring. Thus other labeling and identification systems which would make the task of taking an inventory easier and more efficient were researched.
Barcoding Spine of Books
One obvious solution for making a library inventory more efficient is, in addition to adding a redundant barcode as mentioned above, to place barcodes vertically on each book. Any barcode scanner would be able to read a barcode label on the spine, without the user having to even handle the book in most cases. Library staff who wish to find books which have been misplaced in the studies, for example, would be able to scan in the barcodes of the books on the shelves much more quickly than conventional methods. Scanning in all the books would detect "problem" books which donít belong, and permit the staff member to charge these books out to the study properly.
Barcoding the spines of books is a perfectly feasible system that Widener Library can implement at this time. The cost of barcode labels is relatively cheap (1 cent per label in most cases), and can be easily attached to the spines of books in a similar manner as the call number spine labels are applied. Spine barcoding also will not necessarily require new scanning equipment, as the CCD scanners will certainly be able to read the codes.
There are, of course, many problems associated to attaching barcodes to the spines of books:
When talking to the bindery department of Widener Library, it was discovered that almost all of the bound books have a spine width greater than three tenths of an inch (.3"). Thus, if the barcode width were adjusted accordingly, it could easily fit on almost all of the volumes within the stacks. Only 5% of the holdings, such as magazines and pamphlets, have very thin spines which cannot support a label of any width. It was decided that the original redundant code would suffice for this small percentage of the items.
It is true that the conventional code 39 Widener barcodes would cover approximately two inches of information on the spine. However, a denser symbology such as ëcode128 double densityí would bring the length of the barcode down to approximately an 1.5 inches, without compromising scanning efficiency. Also more aggressive laser scanners could be employed in order to improve the first read rate (FRR) of the barcodes, making the scanning of the spine barcodes easier for the library staff.
Thus, spine barcoding is not as problematic as it first seems, although it still will inevitably cover up book information and possibly look obtrusive to patrons as well as the library staff. In order to alleviate these specific problems, other types of codes were studied.
Two Dimensional Codes on the Spine of Books
Two dimensional symbology is a very attractive labeling scheme which incorporates the advantages of spine barcoding, while eliminating some of its shortcomings. In a nutshell, a two dimensional code is analogous to a barcode, except it lets one encode more information in a smaller amount of space. Essentially, it might be more feasible to use 2-D codes in place of spine barcodes.
Two dimensional codes look like a square or rectangular data matrix holding an array of small black and white squares (pictured above). As opposed to linear, or one dimensional barcodes, two dimensional codes are able to encode data horizontally, vertically, and even diagonally. Not only can the matrix hold more information, but it also allows for a dynamically programmable amount of data correction.
More specifically, a 2-D code is able to encode a 14 digit Widener Barcode in an area of .2" by .2" using a normal 300 DPI laser printer. With a better, higher resolution laser printer (2500 DPI, for example), matrices of less than .03 inches squared holding the same data can be achieved. Clearly, a code of such small magnitude has great importance when applying them to the spines of books. Much less information will be covered by the labels, and they will be much less visually obtrusive to the library patrons.
Another advantage of 2-D symbology is that it incorporates the use of error correction.. While barcodes usually employ a CRC digit to detect bad reads, two dimensional counterparts are not only able to detect the errors, but are able to correct them as well. In other words, when a 2-D label is damaged in some manner, the scanner will employ over ten different error correction algorithms in order to read the code correctly. Studies have shown that up to 50% of the label can be damaged and still be readable by the 2-D scanners. Error correction may have a dramatic effect on spine labeling as labels which are spoiled through normal wear and tear will still be readable by the scanners.
Two dimensional scanners are basically cameras which take a picture of the 2-D label. The label is then read through image processing techniques which are able to reorient and realign the matrix. For these reasons, 2-D readers can scan in matrices from any orientation relative to the scanning surface. The advantage of a 360 degree scanning freedom becomes significant as it has potential to make the scanning of codes a much easier task than with orienting a laser scanner to a barcode, for example.
The advantages of applying 2-D codes to the spines of books are very attractive. The costs of printing labels are comparable to those for barcodes, as they are easily printed with a wide range of printers. The only major disadvantage is that the use of 2-D will require new scanning equipment which will cost substantially more than the current CCD scanners ($1500 for a 2-D reader, $300 for CCD).
It should be noted that much of our research in two dimensional symbology was based on the findings of International Data Matrix. ID Matrix furnished the project with 2-D labels, scanners, and pertinent information concerning the applicability of 2-D codes to the spines of books.
One final identification system that piqued our interest was the ability to be able to communicate books remotely, without having to scan labels through conventional methods.
Radio Frequency Identification Devices
Radio Frequency Identification Devices, or RFID for short, use small tags composed of electronic components in order to encode a relatively large amount of information. These tags can communicate with an RFID transceiver, which basically sends and receives signals to and from the tag. Not only does the transceiver detect the presence of an RFID tag, but it actually reads the tagís unique code in order to determine itís identity. In essence, the transceiver sends a signal to the tag, and the tag responds back by identifying itself.
These devices can play a monumental role in Widener library as a RFID tag may be placed in each volume, giving every book its unique code. Books with RFID tags can be queried from the RFID transceiver, letting the library staff know the location and status of the volume. Using an RFID transceiver wand, or any other type of portable transceiver, would let the user sweep through the stacks, querying each book in order to determine whether they are in the correct location. Inventorying the entire library may turn into a relatively simple task as it would only require someone to walk through the entire stacks and catch misplaced books.
Another application in which RFID can play a substantial role is during check out. Instead of manually scanning each book normally, you could simply sweep the book containing the RFID tag over the transceiver. The transceiver would read the unique book code, and then be able to check out the book to the patron without any complex handling of the volume.
RFID tags are substantially more expensive ($1) than barcode or 2-D codes, but are not necessarily prohibitive considering the cost of putting each book on the shelf, and itís potential benefits for our tracking system. One main problem with RFID is that currently it is hard to distinguish between two or more tags in the same vicinity. In other words, when tags are placed less than 2 inches from each other, the transceiver gets confused between the two signals. Another disadvantage to RFID is its relatively short range of 5 inches.
Despite these downsides, RFID remains as the most ideal solution to our primary problem of continually tracking each book within Widener Library.
Conclusion of Identification and Labeling System
At a bare minimum, we would hope that Widener Library would employ redundant barcodes on the outside cover of the volumes. As outlined before, these identical codes play an integral role to a fast, effective tracking system.
When looking towards inventorying the entire library, or small sections such as the studies, it is paramount to have an identification system in which book information is easily obtained in a rapid manner. Both barcodes and 2-D codes attached to the spine will achieve this ideal to some extent, as it will allow books to be scanned in with a minimal amount of handling. While barcoding has itís disadvantages, it can nonetheless be an effective measure for inventorying. 2-D symbology gives a more attractive alternative to spine barcoding, yet it requires employing a relatively new technology which is not as proven as barcodes.
Finally the prospect of using RFID prepares the library for the most ideal tracking system. Although it requires expensive equipment, and might not allow all of our applications due to its current limitations, it should be noted that RFID is a constantly evolving technology which will inevitably become an increasingly important identification system.
Migration
Although our main research focused on determining an identification and labeling system for Widener Library, it was equally important to come up with an efficient to actually apply these labels or RFID tags to the books. There are mainly three different types of books in Widener Library in which we must consider for the evolution to our ID system: 1) Books which have barcode labels, 2) Books with no labels, but have a bibliographical record in HOLLIS, and 3) Books without any labels or references in HOLLIS.
Since there are already exists a project devoted to entering all of the bib. records into HOLLIS, our main focus was with books which are already labeled. Our migration procedure involves two major plans -- One that will concentrate on books which are currently in the circulation process (i.e., books that are being checked out and returned), and another which involves updating the labels of books which are within the library.
Labeling the books during while the book is circulating will be a fairly straightforward procedure. When a book with an old barcode enters the library, it will immediately enter a "Migration station," which will have the task of printing out and applying new labels on the cover and spine (if applicable). More specifically, the Migration station checks the existing labels of each book. If there is an old barcode, the staff member at the station would scan in the barcode, which would cause a computer to print out two identical codes. These codes would be applied to the front cover and spine. If the book already has new labels, then it will be simply passed along as it already has been migrated.
Migrating book labels within the stacks would be a similar procedure. A portable printer would roam through the stacks in a logical manner, and apply labels just like as before. In essence, there would be a movable migration station that would update old books in the stacks. Both of these procedures require laser printers capable of at least 300 DPI, scanners, and software to print out the new barcodes and/or 2-D codes. An example of the program which would execute the immediate printing of the barcodes is included with this report.
This migration scheme will eventually reach every book in Widener
Library, and thus be an effective means to labeling the volumes.