PromptCat & Bib Notification at UMCP

Presented by John Schalow at the CAPCON Users Meeting on November 22, 1996 in Washington, D.C.

Please note, below are the slides and notes for this presentation transferred from Powerpoint .

University of Maryland System (UMS)

There are 13 libraries on 12 campuses and most UMS campus libraries are joined to our implementation of the CARL integrated library system, which we call VICTOR. Each library has a local database showing their bibliographic holdings and shelf status and besides this, there is also a UMS global database which combines all of our holdings. The global database contains the bib record for the first campus which updates on OCLC and this record is only overlaid if a record comes in with a higher ranking according to an algorhythm in our loader program - mostly based on encoding level, but also if the record is revised LC copy.

UMCP Libraries

There ae 7 libraries on the College Park campus serving 33,000 students and 2,500 faculty. The libraries hold over 2.5 million volumes and nearly 75,000 volumes were added in fiscal year 1995/96. There are 211 library staff. Technical Services consists of Acquisitions, Cataloging, Catalog Management, and Preservation. There are 26 permanent full-time cataloging staff.

PromptCat Objectives

So, why did we begin to use these two OCLC services, PromptCat and Bib notification. With PromptCat, we wanted to bring bibliographic records into VICTOR faster using our existing OCLC loader program to load records into the VICTOR system, both local and global. Any programming changes would have taken awhile. We decided to accept OCLC bibliographic copy at low encoding levels with no editing. The principle which we are following is to get a useful bibliographic record - maybe brief, maybe incomplete CIP - into our system and then complete the record later by replacing it with a bib notification record. Process books faster - handle the books fewer times Acquisitions staff check the book in, apply barcode, write the call number and branch/location in the book, create an item record, and send the book to preservation for labeling, stamping and security stripping. Focus Cataloging staff on uncataloged collections Particularly special collections. And not just backlogs. I recently did an inventory of uncataloged materials throughout the libraries and found lots of work for us to do.

Bib Notification Objectives

Support PromptCat service by automatically upgrading bibliographic records Acquisitions is accepting incomplete CIP records - we are not completing the collation statement nor are we reviewing the record other than to make sure there is a match. Raise quality of records in VICTOR at a reasonable price - actually the price for bib notification right now is free! Which is an incredible bargain! CIP records are being replaced and records with low encoding levels are being automatically replaced.

Implementation steps

We identified the approval plan as the place to start. What we did was identify the publishers and subjects in our approval plan which we never return and we set up a profile for these books which we called �keepers� These books became essentially a �purchase plan� - books which BNA would ship us which we could not return. These 9,000 titles were then designated for PromptCat treatment and we started in September of �96. Profiling decisions with OCLC In order for us to begin our PromptCat implementation quickly, we decided to use existing CARL system software which meant acquiring records the same way as through our daily OCLC load. So acquiring the records via EDX file transfer. We chose not to accept a data record because we have an acquisitions record - and we do not use any catalog cards. We chose to set holdings on OCLC immediately though that has presented some problems. We are sorting the reports by title. We are receiving cataloging for all encoding levels and cataloging sources. We thought, why limit? And if any categories of records turns out to be a problem, we can always change the decision and limit by encoding level or source. Since the purchase plan only ships books, record type has so far only been books. We chose not to add vendor data to the MARC records nor have we asked OCLC to edit the records or add user data though we are still considering having OCLC add a 949 field to generate a holdings record with a temporary location. We are currently loading the records with no item record and manually adding it within days of the load.

Procedures developed

We developed with the Selectors a detailed Branch/location LC class # mapping scheme. So Acquisitions has a chart of LC class numbers mapped to library branch and locations. Since the LC class number is on the BNA slip, they can quickly assign the branch and location based on this chart. Currently, the selector is reviewing the branch and location and changing it in the rare instance this is needed. The selector is also evaluating the accuracy of the BNA purchase plan profile. We plan to eliminate Selector review for most materials. Cataloging staff receive a copy of the BNA slip and the OCLC PromptCat report which identifies titles with series - so far 34% of the copy has a series. They then check numbered series for special treatment - cataloged separately or cataloged together, as well as whether the series is new to us so we can establish it and perform any necessary NACO work. Most of this work is done without the book in hand though the book is nearby and can be retrieved for verification and problem resolution. The book has a slip in it for the Selector to wirte decisions as well as a place for Cataloging to write a call number in cases where a series decision affects the call number. We are also able through this method to catch some multi-volumed monographs. However, I think we have drilled it into everyone by now how important it is to follow LC cataloging practices. Making exceptions simply makes it harder to implement batch processes.

Where we are now

Current accomplishments - Loading records has become routine. Problem resolution has become routine though we still occasionly find something new. Staff have learned who to turn to for resolving specific problems - cataloging and editing - software and the OCLC load - etc. Acquisitions staff have been trained and we will continue to evaluate what further training is required. PromptCat books are getting to the shelves within two weeks as opposed to two months under the former system. The quality of the bibliographic records in Victor remains high. There is less CIP than we expected and it appears to be upgraded in a timely manner by bib notification. There have been no complaints about the quality of PromptCat records from public services.

Selected statistics for PromptCat

We are averaging around 178 titles per week and around 88% of these titles sent by the vendor receive bibliographic records via PromptCat - that�s 948 out of 1,069 in total for the 6 week snapshot or 158 out of 178 for a weekly average. 10 records out of 1.069 were not sent because our symbol was already attached indicating we already own the book. Most of these we have been able to send back to our vendor since the vendor�s own duplicate checking routines have failed. 111 out of 1,069 titles could not be matched with OCLC copy - that�s around 10% - now, when you search these same records in OCLC, some of them have copy, but OCLC chooses to not send the copy if there is not an exact machine match - since we don't have access to the vendor manifest, it is hard for us to figure out why there aren't matches, but my assumption is that it is related to the vendor's manifest info. We have had no problems with vendor control numbers nor have we had errors which OCLC�s report can identify.

Selected statistics for PromptCat

Quality of records is high - 11% of copy is encoding levels 8 & K - 85% of copy is DLC (74% of this is C#P - OCLC�s CIP upgrade service) Now, C#P, this is abit of a concern for us since we thought Bib Notification would automatically send us LC upgrades for CIP. The fact is that the way it is set up now, The OCLC CIP upgrade service upgrades a record and changes the encoding level to blank which makes it appear to be the highest level of copy. When LC reissues the record, it overlays the OCLC record, but because there is no change in the encoding level, it is not sent out via Bib Notification. I have not found this to be a problem, but where we all regard LC copy as the highest level of cataloging, I would prefer to end up with the LC record. Screening steps are still necessary as I said before - 34% of the copy has a series. We need to identify books in series which are classed together. We also need to identify series which are new to our collections since we also create a serial record for the title with a note indicating that the series is either classed together or separately. As a NACO library, we also perform series authority work for series which are new to our collections. We have been using the PromptCat report and the BNA approval slips - which both identify titles with series - to identify series which require special treatment. A few titles require cuttering work and 6 have been completely wrong records. During this initial training period, any and all exceptions are being given to Cataloging for resolution and we will determine what additional training is required of Acquisitions. The numbers of exceptions have been small.

Still to be accomplished

Load records sooner - currently taking about 2 weeks, need to bring it down to 1 week. We need to wean the selectors from looking at every book. We are currently evaluating our call number branch/location mapping scheme & identifying areas which need continued selector review. There are some catagories of books which Selectors will probably always have to review to assign the proper location - Reference materials is the biggest. Change procedures to take full advantage of PC service. It is so important to look at what PromptCat can do and try to implement it to its full advantages rather than adapting it to suit your current procedures. At College Park, we need to continue to review our local practices in light of PromptCat - we need to change loader programming to accomodate acquisitions and bibliographic data in one record, we need to evaluate the usefulness of some of our local cataloging practices. And see if we need to make changes - while adhering to the principle - do not hold up the book. Also, continue to train Acquisitions and copy cataloging staff to perform routine editing and cuttering so that fewer books are bounced to the original catalogers. We have explored the option of shelf-ready processing by our vendor BNA. We developed specifications and BNA made a cost proposal which we subsequently rejected for two resons: 1) BNA needed more time to perform the work than we wanted to give them. 2) We asked for things which required programming and we didn�t want to pay that much which was from $1.47-$1.80 depending upon what was done.

Bib Notification

The Bib Notification service delivers around 20 records per day into an EDX file 7 days a week. We have chosen to batch these files and load them once a week. Our OCLC loader software provides a detailed report showing how many records were loaded and how many overlaid existing records. Unfortunately, our loader does not match against 019 fields - that�s obsolete OCLC control numbers. It only matches on the primary OCLC control number, so we generally have several non-matches and have to manually go in and delete the old record and move its holdings to the new record. We need to update our loader! Other reasons for no matches have been that we haven�t always removed our holdings from OCLC when we don�t own it - this has been rare for us fortunately. We have a complex overlay hierarchy which has DLC records at the top - and if the encoding level is the same or higher, the incoming record does not load. So some of our bib notification records are not loading - because we raised the encoding level when we brought the record into our system the first time. For the global database, another campus may have brought the record in at a later time when the encoding level was higher.

Selected statistics for Bib Notification

For the 10 weeks ending 10/16/96, the quality of the incoming records was high. Fully half of the incoming records - 479 - were Library of Congress cataloging. Unfortunately, 16% or 184 records were UMC records - records which our own catalogers had upgraded. GPO was small, only 5, but I tracked them because we are a 100% govt doc. depository and we receive all GPO records through OCLC�s Gov Doc subscription service. So we got those records twice. If you subtract the 38 C#P (OCLC CIP upgrade service) records from the Other category, and consider them DLC records, which they are, the DLC count goes up. Average weekly receipts has increased from the 1st five weeks to the 2nd five weeks, probably due to a combination of slow cataloging or publishing activity in the summer and then our implementation of PromptCat in late September. We are currently averaging around 140 records per week.

Recommendations to OCLC PromptCat

Recommendations to OCLC for PromptCat we want more options for setting of holdings. RIght now, you have two options, you can set your holdings immediately or in 21 days. For us, since we selected immediate setting of holdings, this means setting our holdings on Saturday or Sunday. This is too soon, but 21 days is way too late. So, how about 7 days or 10 days? Bib Notification - limit records by examining the 040 string - so if say, UMC - our OCLC symbol - is the last symbol in the string, do not send the record. Around 16% of the current Bib Notification receipts are our own UMC upgrades. Another example of using the 040, if C#P is in the string, ship the record when DLC reissues the record. There were also some GPO records which we would subsequently receive on our monthly OCLC Gov Doc tape load. So we need to eliminate the overlaps in the service. Also, I could see where some libraries may want to control what formats are received through Bib Notification. We are currently looking at records for non-book formats, but the numbers of records for other formats is small and at this time, I see no reason to limit by format. In conclusion, these OCLC services are important tools for us to use in the continued automation of cataloging processes.

If you have comments or suggestions, email me at js368@umail.umd.edu