Ex Libris MELvyl Tells (EL Mel Tells)!

Volume 1, Number 3, September 2002

In this issue:

• Merging Records in the Melvyl Union Catalog
• Options for Title Searching in Melvyl-T
• CDL Staff Profiles

Did I say, see the Melvyl-T prototype catalog? http://mel-t.cdlib.org

Merging Records in the MELVYL Union Catalog

Overview

The MELVYL® catalog is a union catalog, which means that it combines records for materials from all of the University of California campuses into a single database. Naturally, many campuses have purchased the same books or maps or sound recordings so there is some duplication of material in the catalog. Rather than force users to look through lists that have duplicate entries the software merges together records for the same items so that the data about the item only appears once with all of the related campus location information listed on the screen with it.

Although the user sees only one record with multiple holdings each record submitted by each library is stored in its entirety in the database.

Figure 1 - Campus records are input to MELVYL

As records enter the database they are grouped with other records for the same work through a complex algorithm that determines "like" records in the database. Records are not merged, however. Each record is in the database and is indexed and retrieved on its own. No merging takes place until the time of display. When records are retrieved and before they are displayed on the screen, each record in the retrieved set is used to also retrieve all of its merged "partner" records.

Figure 2 - Search and merge for display overview

Once a group of records is retrieved and needs to be displayed, a display record is created using the bibliographic fields from one of the records and incorporating the location and holdings information from all of the records in the retrieved set. This combined record is then used to form the MELVYL display.

Figure 3 - Group of "like" records

The display record is chosen through a simple point system: records are given one point for each field that is indexed, plus five points for a table of contents field (MARC 505 tag). In this way the set of bibliographic fields that are shown to the user are from the most complete record in the set. As records are updated in the database the selection of the display record can change based on the points.

Figure 4 - Combined record in MELVYL display

You can see the individual records submitted by each library by clicking on the linked campus name after Library at the bottom of the bibliographic record in the long display.

Library UC Irvine UC Davis UC Santa Cruz UC Santa Barbara UC Santa Barbara UC Berkeley UC San Diego All

---

Options for Title Searching in Melvyl-T

A title is a title, is a title—not! In Melvyl-T, there are several options for title searching. They are:
Title words
Exact title
Journal title words
Exact journal title
Main title words
Exact main title
And even
Author/title

In Basic Search, a user can search by title words, exact title, journal title words and exact journal title. You can search by these and Main title words, Exact main title and Author/title in Advanced Search, and, of course, by all of these indexes in Command search as well. You can also Browse in the title fields.

Now let’s figure out how and when you’d want to use all of these.

Title words are words drawn from a variety of indexes, including title, main title, series title, and conference title, and you would use this field when you know some words out of a title, but aren’t sure of the entire title. Title words can be searched in any order. For example, you know a professor mentioned a book that had the words feminist, black, and white in it, but you can’t remember the exact title. So choosing the title word index from the dropdown box, type in feminist black white and you come up with two editions (in the Melvyl-T prototype) of the book Common differences: conflicts in black and white feminist perspectives.

The MARC fields that are searched for this search are the following: 100, 110, 130, 210, 212, 214, 222, 240, 241, 242, 243, 245, 246, 247, 400, 411, 440, 490, 700, 710, 730, 740, 780, 785, 79X, 800, 810, 811, 830, 840, 89X.

When searching for Exact title, word order does matter. Searches must be left-to-right starting at the first word, omitting the initial article. When you do a search here, you are searching words from all forms of titles in the bibliographic record, including main entry titles, added entry titles, series titles, series added entry titles, abbreviated titles, subsequent titles, etc. The same MARC fields are searched as when you do a title word search. You can use truncation when searching in Exact title to finish off your title, for example, children of a lesser* pulls up the title Children of a lesser God.

Journal title words searches these MARC fields: 100, 110, 111, 130, 210, 212, 222, 240, 241, 242, 243, 245, 246, 247, 700, 710, 730, 79X.  You can use this index to find all the journals with the words cell biology anywhere in them.

Whereas, if you’re looking for the Exact journal title Cell biology and toxicology you’ll find that your words must be left-anchored, with words in order starting with the first word, omitting initial articles, and using the truncation symbol if you wish to truncate your title. Exact journal title searches the same MARC fields as above.

Now what exactly are Main title words? Main title words are the title of the work itself, not other titles that can appear in the records such as series titles, titles on added entries, etc. The MARC fields searched include 130 or 240, 245 and 247. If you do a search for “Hamlet prince of Denmark” you will get the Shakespearean play by that name, but not a translation into Chinese that has the English title as an added entry. Main title words and exact main title are only available from the Advanced search screen or in Command search.

Exact main title is essentially the same as main title words, but in this index, you must begin your search from the left-most character, eliminating leading articles. Because this index does not include series titles, or a variety of other title added entries, it retrieves smaller results sets than the Title Words index.

In the Melvyl-T prototype file, when you search for title words, drug abuse, you pull up 1,417 items. Using exact title yields 301 items; exact main title pulls up 229 items.

Author/title draws from the following MARC fields: combinations of 100 (author) or 110 (corporate author) fields, and their related titles, 240 (uniform title) or 245 (main title). Added entry fields with 700, 710, 79X and $t (title) subfields are also included. This index allows you to search for author and title in the same search. For example, you could search for twain huckleberry or hawking brief history.

More than you want to know about title searching? All of this information will be available from an Expert Searching section of Melvyl-T help, so don’t despair.

These indexes can also be browsed, but that’s a topic for another day.

---

CDL Staff Profiles

Continuing on with profiles of those who are helping to build Melvyl-T, meet the following people:

Anita Auyang
	Annita Auyang, our Oracle database administrator, came to us from the private sector where she has worked for Charles Schwab, PeopleSoft, and British Petroleum. She started at the CDL this past July. Oracle is the software that allows us to construct the Melvyl-T database that houses all our bibliographic data--24 million records! It is the back-end to the new Ex Libris Aleph 500 Web-accessible application. As the Oracle database administrator, Annita creates and maintains the Oracle databases and configures the Oracle environment on the Unix machine for best performance.
She also configures the backup for all-important recovery purposes, and assists people in querying the database. Annita comes to us from the private sector, so brings a unique viewpoint. She is struck by the support, friendliness and family-feeling at the CDL, and was positively impressed--and a bit surprised--with the up-to-date equipment and computer support at the University. Annita enjoys spending time with her family, and with interests such as flower arranging and swimming when not at the CDL.

Randy Lai
	When I talked to Randy, he was very pleased to report that a milestone had been reached in the work with which he's been involved--the loading of the Melvyl-T production database. Randy works with colleagues Lynne Cameron and Mark Reyes, helping to set up and run the input files from the campuses (campus snapshots), as well as from the other streams. When the snapshots first arrive, they go to a Virtual Host, are then archived using Tivoli Storage Management (TSM), and then moved onto one of the different servers that have been used during the building of the Melvyl-T catalog: Pinatubo, Hermes, and Mercury.
Randy also has learned how to use the job scheduler (“a system designed to allow unattended operations, automate operator functions, and control report distribution”) to automate some of the functions that have been done manually at first. He has also helped run the snapshots through the converter programs, checking the log files for statistics and error messages, and reporting these. Who would have guessed that, until recently, Randy played tennis at 6:30 A.M before coming to work? And that he's mastering the art of the Chinese flute?

Sean O'Hara
	Sean O'Hara, entering his 25 th year working for the University of California, is now the remaining database administrator for legacy Melvyl. (Thanks, Sean, for keeping this old faithful running!) For Melvyl-T he plays several roles: one is as a member of the CDL statistics team, capturing data in the new system, helping to make the statistical reports “look good” and when necessary, running data through software to generate the kinds of reports we want. (He has maintained legacy Melvyl statistics for some time.) He has played an analyst role (one of his strengths) during different phases of the Melvyl-T project, including in helping to set standards for the new Melvyl job scheduler.
He has augmented his database administration experience with training in Oracle in order to assume a role as a backup to Annita (see article above). In his non-CDL life, Sean is Chair of the northern California branch of the Mediterranean Garden Society (http://www.mediterraneangardensociety.org/ ). [See the September issue of Sunset for yet more information on this group.] While some colleagues are off at ALA, DLF and other library-related conferences, Sean is busy speaking and arranging conferences and symposia connected to his position in the Mediterranean Garden Society. A perk of his position, and passion for gardening, has been the opportunity to travel and see beautiful gardens throughout the Mediterranean world. A related perk for us at the CDL is the plethora of wonderful plants we have in our office, flourishing under the care Sean's very green thumb.

Felicia Poe
	Felicia Poe joined the CDL in December 2001 as a Service Design Analyst in the Digital Library Services Group. She has played a lead role in designing the functionality of Melvyl-T's user interface. She analyzes the software, as presented by the vendor, modifies it when desirable and possible, and integrates it into our vision of how we want Melvyl-T to look, taking into account guiding principles. These principles are based on input from librarians, end users and basic usability principles.
Felicia has also worked extensively on the Melvyl-T Help, particularly the Expert User Help, building on the contributions of Ann Jensen, her predecessor at CDL. Her work as a law librarian gave her broad experience in information retrieval. The skills she gained as project manager and information architect for the Indiana University Web Team (where she led the redesign of the IU home page and admissions sites) contributed to her knowledge of research in the higher education environment. Combining her past experiences in different settings helps her approach her work at the CDL. When Felicia left the Bay Area to work on her Master of Information Science degree at Indiana University, she had no idea what “Big Ten” (where sports are everything!) meant. She is very happy to be resettled in San Francisco.

Virginia Sinclair
	Let me see if I can properly explain what Virginia Sinclair does for the Melvyl-T project. Virginia coordinated and monitored the “unloads” from the 23 sources in legacy Melvyl that are not sending their records into Melvyl-T as “snapshots”, as the main UC campuses are. Examples of these institutions are the UC Davis Law Library, the affiliated libraries at UC Berkeley, the Center for Research Libraries. Once these records have been unloaded or pulled off legacy Melvyl, they are then run through converter programs that put them into the form that the Aleph 500 catalog needs, and loaded into Melvyl-T.
For example, records that previously had holdings information in the 920, 930, or 950 fields now have to have that information written to the 852 fields. Virginia also creates benchmark files for all sources going into Melvyl-T to test for the myriad of possibilities of how records may appear coming into the CDL, and how they must look in the new system. She checks the output after conversion to ensure records emerge as they should—an iterative process. Benchmarking has always been conducted on legacy Melvyl. In the future, Virginia will be working on Melvyl-T's merge algorithm. Virginia practiced many of her programming skills working on legacy Melvyl, writing programs in PL/1. (Anyone else out there remember that?) Her passions include the Edinburgh Festival, an over 50-year old celebration of the arts (and including the Fringe Festival), where she's been a nearly annual attendee since 1991, and reading and collecting medieval mysteries.

Margery Tibbetts

"Margery knows everything!" This is something we have said frequently while making CDL road and plane trips to the nine campuses with Margery. Wonder what kind of truck is passing on the interstate? Ask Margery. Have a question about intricate beading? Ask Margery. She always knows the answer. I attribute Margery's broad and deep knowledge to her roots as a fourth generation librarian. (Ask her about that sometime!) Among many other things at the CDL, Margery has primary responsibility for the Z39.50 server and how it interacts with the general world of Z39.50 clients, with Request, so that interlibrary loan requests are made correctly, and with the campus circ systems so that campus circulation information is retrieved and displayed properly. She is, in fact, the Project Manager, for the Z39.50 server developed by the CDL, the backbone of the Melvyl Web system. Another active project for her right now is also implementation of SFX/UC-eLinks for UC. Anyone who knows Margery in the slightest knows that the she is a polymath; in addition to her many "at work" talents, she is also a craftswoman--an award-winning quilter, who also enjoys beading and needlework.

Tell us what topics you want to see covered here. CDL Feedback