Present: Bob Alan (HOTS); Pat French (UCD); Crystal Graham (UCSD); Lynne 
Hayman (UCSB); Carol Hixson (UCLA); Rebecca Utz (recorder); Brian Warling 
(CDL) 
  

Minutes were approved pending changes regarding the discussion of Decision Point 2. 

Discussion of the minutes led to another discussion of Decision Point 2 
(Paper version changes title requiring successive entries; ejournal at one 
site).  There is a difference of opinion on whether a new record should 
be created for the electronic site when the paper journal changes title. 
The discussion was tabled to be addressed again at a future date. 

Prior to the conference call, French suggested that the central cataloging 
agency might be able to supply both a record for the electronic version as 
well as a composite record, in order to produce a product all campuses can 
use. The premise would be that composite (i.e., single) records would 
be added to Melvyl, but that campuses could elect to use the other form 
locally if that was their decision.  There are records for most of the 
commercial electronic journals in OCLC'S Intercat Database already. 
French's idea was that these records could simply be supplied as an 
alternative without doing any work on them.  Graham stressed that standard 
copy cataloging work would also be needed, including authority work. 
UCSD's experience is that many of these records need substantial copy 
cataloging. Graham stated that in practical terms, asking the central 
cataloging agency to distribute both forms of the record to provide a 
local option would mean duplicate cataloging in most cases and this is not 
desirable, nor can it probably be justified on a cost basis.  Hayman 
stated that UCSB, as well as other libraries in the system, including 
Berkeley will not be able to utilize single records because it would not 
be possible to define a matching data element that would enable 
overwriting of print records.  The distribution of separate records, in 
addition to records representing print, provides each 
campus with something it can load and use, even if editing is required. 
Distributing both records means each campus can choose to load one or 
both, as one record does not logically preclude the other.  Graham 
stated that this would not be cost effective, Alan stated that it may or 
may not be cost effective. Hayman stated she felt it would be premature to 
reject the proposal on the basis of cost prior to further consideration 
and study, particularly since this may be the only means of providing 
several or more campuses with a product they can use. 

The group agreed to meet from 9-11 rather than 9-12 on January 22. 

French reported that the Dec. 18 minutes will be on the website soon, 
and that she will update the report outline on the website. 

SOPAG has just released a new planning document called "Draft 
Requirements Analysis for the CDL Public Access Catalog", written by Mike 
Berger and Karen Coyle of CDL Technologies.  SOPAG is distributing this 
document widely and requests comment by Jan. 28. 

ACTION: Alan will forward a copy of this document to the task force. 

French reported that on Jan. 13, she and Hayman spoke to Cynthia Clark, 
who is now TFER's formal liason with SOPAG.  They asked Clark to fill them 
in on SOPAG's discussion at their meeting Friday, Jan. 8 on task force 
progress, and they filled Clark in on the status of TFER's work to date. 

During the discussion, Clark reported that Cecily Johns (UCSB) reported to 
SOPAG on Jan. 8 that TFER has been working hard to reach a concensus on 
the overall cataloging approach but that the variety of opinions within 
our group has made it difficult for us to agree on what the guidelines 
should be.  SOPAG then held a discussion of what to do about the impasse 
the task force has reached.  There were suggestions that perhaps we should 
present a modified initial report which outlines the cataloging approach 
being used nationally (CONSER, LC, GPO) so that SOPAG could assess the 
situation and decide which approach to adopt.  There was also discussion 
about the desirability of having more public services input into this 
decision.  The suggestion was made that this be added after our report is 
received rather than asking us to build in a process for more direct input 
up front. 

Hayman and French responded that we do not feel that we have reached an 
impasse, but are on the way to defining the principles and practices which 
would underline cataloging guidelines provided compromise can be made on 
decision points and provision of cataloging that results in a product all 
of the campuses can utilize. Although it is challenging, progress is being 
made.  Hayman and French recommended that the task force continue towards 
completing the original charge of writing a guidelines document and Clark 
agreed that this is what we should do. Clark will report this back to 
SOPAG. 

They also talked about the great amount of detail involved with 
fully completing our charge and the fact that we may need to deliver a 
report on Feb. 15 that does not fully account for all of the technical 
details. Clark indicated that SOPAG's main interest is 
in knowing as soon as possible what the cataloging guidelines for Tier 1 
and Tier 2 titles will be, what the records will look like and contain and 
how we propose sharing them with campuses.  She agreed that if we 
concentrate first on writing a document that outlines the cataloging 
specifics (content, tagging, application), we can present options for 
record sharing/loading in more general descriptive terms.  The technical 
details can follow as we are able to work them out.  If we can identify 
the cataloging guidelines and conceptual basis for record sharing and 
linking with the CDL Directory (i.e. present the "vision") by the 
deadline, we can continue working on the details as a next step. 

The guidelines document should be as complete as possible for the 
cataloging guidelines, but it may just identify other areas that need more 
work or time (such as policy for analysis, plans for other types of 
materials, and options for linking with the CDL Directory). 

A.  One of our goals is to produce a guidelines document by Feb. 15. This 
document will cover record content and tagging specifications.  It will 
also include guidelines for deciding which type of record to use in 
specific situations. 
        We will adopt the following strategy to reach this goal: 
        1. Review and revise existing guidelines documents. 
        2. Use UCSD's single record guidelines and Nancy Olson's 
                Cataloging Internet Resources guide. 
        3. Identify things to add, such as: location, reporting schemes, 
                call no. display, holdings display. 
        4. Provide sample records. 

ACTION: It was proposed that French and Graham work on this.  Graham will 
take the lead in guiding the group through this process. 

ACTION: Graham will also provide some sample records to the task force. 

ACTION: Hayman volunteered to draft the narrative report and share this with task 
force members for input. 

B. A second goal is deciding on  and adding application guidelines (i.e. 
identified decision points).  We will continue working through the 
remaining decision points and also reach a working definition of 
equivalent content.  Decisions are also needed on whether or how to 
integrate electronic resources with records for other formats when they 
exist (i.e., CD-ROM, micro). 

ACTION: The group will focus on these issues at the Jan. 22 meeting. 

C. A third goal is to identify items of the charge to be addressed beyond 
the task force's current deadlines.  These may include: 
        1. Further study and input on analyzing individual titles 
contained in large databases. 
        2. Expanding guidelines to include other types of materials (e.g., 
archival finding aids, local websites, etc.). 

D. A fourth goal is to identify options for linking with the CDL 
Directory.  The interview with Brian Warling at this meeting (below) 
identified some possibilities. 

ACTION: Hixson will take the lead on investigating linking options. 

E. A fifth goal is to identify options and remaining work regarding 
record sharing by Mar. 1999. 

ACTION: Alan and Doherty are working on this. 

 Brian Warling, CDL, Web Design Coordinator, joined the meeting by 
 telephone at 10:00. The task force had sent him a list of topics and 
 questions outlining the types of information we hoped we could learn from 
 him.  We started by asking Brian to give an overview of the CDL Directory 
 concept and how it currently functions.  These minutes record the 
 information that was shared in our conversation. 

 The CDL Directory ws initiated soon after the CDL organization was formed. 
 Its immediate purpose was to develop an access mechanism which allowed 
 users to locate CDL resources quickly by title and by broad subject 
 areas.  The CDL developed the design for the Directory and iniitally 
 contracted with an information architecture firm to develop its 
 architecture; CDL now entirely manages the directory. 

 The Directory was conceived as a retrieval tool for CDL licensed digital 
 collections and services, including electronic journals, bibliographic 
 databases and OAC finding aids.  The ability to identify digital materials 
 in a general subject field and to determine which resources were 
 accessible from any particular campus are functions that had become 
 increasingly important and which the current Melvyl catalog in its present 
 form cannot provide.  The Directory was therefore designed to supplement 
 access available through Melvyl but was not intended to replicate it. 
 Subject access available through LCSH and MESH headings in Melvyl 
 provide a more specific, granular subject organization to materials but do 
 not provide the ability to browse in broad topical areas.  This function 
 was highly sought for CDL digital materials. 

 There was also recognition that there are many web pages being created by 
 librarians throughout the UC system and that maintaining these web pages 
 is becoming increasingly burdensome.  CDL desired to develop a way that 
 web access to digital materials shared in common could be maintained 
 cooperatively.  The Directory was conceived as a dymanic site which could 
 be presented with differing iinterfaces tailored to present particular 
 campus views as needed.  Some campus libraries have already expressed 
 interest in using the CDL Directory in place of their own web sites and 
 adding their own customized lists of resources to it for presentation at 
 their campus.  Brian has talked with UCSF, the UCLA Biomedical Library and 
 UCB's Biology Library about using the Directory this way. 

 The Directory was started by building a database of text-based records 
 which described CDL materials.  The records are tagged text files 
 containing data which describes each digital resource (e.g., title, 
 coverage/contents) and access information (e.g., URL and a campus access 
 byte).  The resource description format that defines the records was 
 developed locally with influence from the Dublin Core format.  These files 
 are currently maintained by CDL staff, however CDL wants to develop a 
 different, more efficient mechanism to build and maintain the database in 
 the future. 

 The general record format of the text files can be viewed directly at: 
 http://merlin.ucop.edu/~bnw/ejrecords 

 The two types of data currently found in the Directory records which are 
 unique and highly desired are access information and coverage.  The access 
 information is controlled by a field containing a byte tagged for each 
 campus.  The value is either "0" or "1" depending on whether that campus 
 has access to the resource. This allows the access information to be 
 updated simply by changing this byte if the campus adds or drops their 
 support for the license.  This field also controls whether the title is 
 presented in a campus view of available resources.  The CDL would like to 
 establish a system whereby campuses themselves could update this access 
 byte and turn the responsibility over to them. 

 The coverage information is a statement of what's available online at the 
 electronic site.  The CDL took much of their coverage data from UCSD's 
 MARC records (856 $z, "Materials Specified Note".  NB: UCSD uses $z for 
 this note because their Innovative Interfaces library system does not yet 
 support 856 $3.  This is where the coverage information would be found in 
 records from non-Innovative libraries.).  Coding coverage information 
 (also knows as holdings data) in 856 $3 was recommended by the TFER report 
 in March 1998 and some campuses have been using it since the TFER report 
 was released. 

 In its desire to estblish a more dynamic system for creating and 
 maintaining Directory records, the CDL is interested in exploring using 
 data from cataloging records to supply Directory data.  The primary data 
 types needed include: 
         -description 
         -holdings 
         -URL (or PURL) 
         -subject browsing terms 
         -campus access information 
         [add: -format] 

 Much of this data is contained in a completed MARC record.  New data 
 fields or some other mechanism would need to be established for the campus 
 access information.  The best way to integrate the CDL subject browsing 
 vocabulary would also need to be studied.  The hierarchical nature of the 
 CDL vocabulary carries different subject searching implications in Melvyl 
 (and other OPACs) whose subject indexes are built on non-hierarchical 
 vocabularies.  One possibility might be to add the CDL vocabulary in a 
 specially-defined MARC field which may or may not be indexed in Melvyl. 
 It may be possible to accomodate the hierarchy through the use of 
 indicators or subfields. 

 The best way to link MARC records and the CDL directory will depend on how 
 the data is to be used by CDL.  Three theoretical possibilities were 
 identified, all of which need further study: 

 1.  CDL could continue to maintain a physical database of records 
 describing CDL resources.  Database records could be supplied and 
 maintained by transferring data from MARC cataloging into mapped CDL 
 fields.  A data transfer capability from the CDL centralized cataloging 
 agency to the CDL directory would need to be established.  A unique 
 matching point or algorithm would also be needed if cataloging records 
 are to be merged with existing Directory text files. (If records in the 
 Directory database were reestablished from scratch with cataloging data, 
 matching might not be an issue.)   CDL Directory searching would continue 
 to be based on Directory records themselves.  Under this approach, the 
 CDL Directory database could continue to use an access control field and 
 non-CDL agencies could conceivably have authorization to update it.  The 
 CDL vocabulary could remain based in the CDL database also. 

 2.  Retrieval in the CDL Directory could be based on launching a dynamic 
 search against cataloging records from the centralized cataloging agency 
 in Melvyl.  Under this approach, there would be no physical CDL Directory 
 database for cataloged materials.  Directory search functions would be 
 defined in terms of available MARC data.  There are a number of projects 
 around the country which have linked their catalogs and websites in this 
 way, notably UCSD and Iowa State University.  Once built, there would be 
 no need for regular data transfer or maintenance; the records would be 
 automatically updated through the cataloging maintenance function. 
 There would also be no need for a unique matching number.  This method 
 would require further research to develop a method for maintaining access 
 data. 

 3.  A third longterm possibility is that the CDL Directory may no longer 
 be needed in the future if some of its supplementary functions are added 
 to the next generation UC union catalog. 

 Regardless of which linking field mechanism is selected, there is a need 
 to establish a more efficient process for maintaining URLs in database 
 records.  The task force is interested in a centralized PURL resolution 
 service through which URLs can be cooperatively maintained.  Brian said 
 that he felt this could be a good idea for Directory records also, 
 especially if the PURL server were located at the CDL.  CDL might want 
 the PURL server to be maintained cooperatively by contributing libraries. 
 A PURL server housed at CDL could also supply various electronic site use 
 statistics which would be very useful.  CDL is also investigating other 
 means of monitoring access and use such as certificates. 

 Task force members found this exchange with Brian to be very informative 
 and thought-provoking.  Carol Hixson agreed to take the lead in further 
 investigating options for linking MARC records with the CDL Directory and 
 will be in contact with Brian in the future.