Cataloging Issues
Home

ONLINE LIBRARY CATALOG DESIGN

“While it is true that [catalog] users do not want to end their research with ‘everything’ in the library, it remains true that they do wish to

start with the best possible retrieval pool, from which they then make their selection.” The danger to be averted…is that users “who

retrieve ‘something’ very often miss most of what is actually available without realizing that they’ve missed anything at all.”

--Thomas Mann, a Library of Congress reference librarian, as quoted in Yee (1998)

The physical card catalogs of the past did not present many of the problems now faced by users of online catalogs, and the vendors and librarians who design them. At the time, bibliographic records, authority records and cross-references were interfiled. Full browsing was unimpeded. Today, as users search online catalogs, users don’t find other patrons impeding their progress, but sadly, that simple problem has been replaced by a long list of design and function issues that frustrate everyone in their attempts to locate all materials within the catalog that are relevant to any particular search. Initial automation failed to preserve the integration of the various files. Even for users familiar with cataloging structure, a search of any complexity often necessitates multiple steps to obtain all records. For the majority of users, unfamiliar with library structure and controlled vocabulary, it is often difficult to obtain any relevant records, much less all of them.

Some of the more recent automation products have begun to address integration of files, clarity and variety of search functions and results presentation available, and other issues affecting ease of use and design appeal. When users are asked to describe the look and features they would like to see in an online catalog, they point to Amazon.com. They want natural language capabilities, GUI interfaces, content rich results, and attractive design “like a regular website.” (Puacz, 2002) In choosing three academic catalogs for this analysis, my first impulse was to choose the ones that went for this “look” and content. However, reviewing multiple vendor products by searching the academic catalogs that use them, it became clear that the products that have done the most to improve content and appeal, are not the ones that have the highest degree of improved function. It is the issues of function that I consider to be the most important, and as a result, they are primary in my choice of three library catalogs and the features reported. The vendor products analyzed in the following graphical presentation are ExLibris’s Aleph as represented by the library catalog of Brandeis University (http://library.brandeis.edu), Endeavor Automation Systems as represented by the catalog of the Illinois State University library (http://eisweb1.aiss.uic.edu/isu/cgi-bin/Pwebrecon.cgi?DB=local&PAGE=Firsty), and Carl.Solution as represented by the online catalog at the University of Maryland, College Park (http://victorweb.lib.umd.edu/). Discussion of increased metasearch capabilities, and purely cosmetic enhancements can be found in my textual comparison that follows. There I use examples of other library automation products and online catalogs, which in most cases, better illustrate progress in those areas.

First, looking at clarity of the information and search capabilities provided by the initial search screen presented to the user, and the default search it offers, we can begin to understand the problems of average users, most of whom, as Christine Borgman (1996, as quoted in Horn, 2002) pointed out, remain “perpetual novices.” They are unlikely to find a large percentage of the records relevant to their search in the very few steps they are willing to undertake before giving up, deciding the library has little or no relevant material, and heading for a general Internet search engine, if indeed they have not already done so, bypassing library catalogs entirely.

INITIAL SEARCH SCREEN, DEFAULT SEARCH, OTHER SEARCHES OFFERED

AUTOMATION PRODUCT AND UNIVERSITY LIBRARY OPAC DESCRIBED

FEATURE	ExLibris Brandeis	Endeavor Illinois State Univ	Carl.Solution U. of Maryland, CP
Search Term Box Offered	Yes, 1	Yes, 1	Yes, 1
Default Search Offered	Keyword Any	Any Word Any	Keyword T,A,S
Files Indexed in Default Search	Auth, Bib, Cross	really anywhere	Bib only
# Other Searches Initial Page	6	7	7, pull down menu
Searches Defined?	Yes	Yes	Partial
Search Tips Onscreen	Truncation hint	Yes, extensive	No
Link to Search Tips or Help	Yes, extensive	Yes	Yes
Link to Guided Advanced Search	Yes, 1	Yes, 1	Yes 2, no guide
Link to Other Searches, #	No	No	No
Total # of Searches Offered	8	9	10
Personal Preferences Link	rec # 20-50, content	rec #10-50 on screen	No
Limits Offered Onscreen	None	Last 5 yrs, format/5	None
Limits Offered in Linked	None	F/17, Lang+, Date, Loc	None
Limits Offered in Adv Search	Format, Lang, Date	same as above	Format, Lang, Date
File Browsing Capability Clear	No	Yes	Yes

Next we look at the features offered or re-offered on the results page to understand the breadth of record information offered, quality of navigation, expansion help available for users/searches with too few results, and limits offered for users/searches with too many results. For this Table the searches were “Solar Energy” and “Soalr Energy”

RESULTS : SEARCHING DEFAULT WITH STANDARD TERM, RANGE OF RESULTS #, LIMITS, CONTENT, LINKS

AUTOMATION PRODUCT AND UNIVERSITY LIBRARY OPAC DESCRIBED

FEATURE	ExLibris Brandeis Univ.	Endeavor Illinois State Univ.	Carl.Solution U. of Maryland, CP
Results Appear Quickly	Yes	Yes	Yes
# Results	61	8026	1309
Appear Relevant	Yes	Yes, but way too many	Yes, but too many
Results With Misspelling	No, pop hint	Yes, although sort change	No
Search Terms/Type Report	Not Consis.	Yes	Yes
Navigation Offered By Page	Yes	Only Some	Yes,
Record Items Linked	item #, instruc	#, Title, Availability	#, Title
Default Includes Full Title	Yes	Yes	First 48
Default Includes All 245 $c	Main Entry	Some Records Only	Main Entry only
Default Includes Date	Yes	Yes	Yes
Default Includes Format	Hint in location	Yes	No
Default Includes Call Number	Yes	Yes	No
Default Includes Build/Loc	Yes	Yes	Yes
Default Includes Availability	Yes	Yes	No
Record Changes Available	Yes, custom	None	No
Default Record Sort	date, then author	Relevance	Date, newest
Other Sort Available	Title, Author, Call #	Title, Author, Date either oldest or newest	Title, Author, Call # but last not onscreen
Preferences Available	Link, rec #3-20	rec # 20-50	None
Limits Available Onscreen	None	Last 5 yrs, format/5 except Browse	None
Limits Available Linked	F/8, Lang/12, Date, Bldg/Collec	F/17, Lang(tons), Date, Loc except Browse	F/6, Lang/7, Date
Modifications Linked	Add Term , Choose Boolean, Choose Field/5
Buttons: New Search, History, Marked Records, Union Catalog	Basic/Adv/Modify Basket, No Union	All, although Marked Record Button=“Titles”	Yes, can also mark some or all
Buttons For Help, Acct, Request	All, Reserves	All	No Acct, Request
Print/Download/Email Assist	No-Frames	All
Search Box Avail	No	Full bib lists only	Yes, but at bottom
Query Remain Browser Back	No	Yes, New Search no	Yes
Reset Button	Initial Page Only	Yes	No
Timeout Problems	Yes	Yes	Yes
Server Problems	Lots of no/avail	No	Slow at times

The data available in the full bibliographic record for each item can also be very important to the user. In all three cases this was linked from the results page, although again, in the case of Brandeis and Illinois State, a great deal of information was available without leaving the main results page. In all catalogs, newer records were more complete than older ones. The following table presents the fields reported on the item record for Wartime Writings 1939-1944, translated by Jack Purcell, 1986 if available. The University of Maryland, College Park had no new items and whether a particular field is linked from the record

NEWER FULL BIBLIOGRAPHIC RECORD DEPTH AND LINKS

AUTOMATION PRODUCT AND UNIVERSITY LIBRARY CATALOG DESCRIBED

FIELD, LINKAGE (L)	ExLibris Brandeis Univ.	Endeavor Illinois State Univ.	Carl.Solution U. of Maryland, CP
Main Entry	2p L,Yes	L Called Author	L Called Author
Author, ed, trans	No	With Title	No
Uniform Title	2p L,Yes	L	Yes
Title	Yes	Yes	Yes
Edition	Yes	Yes	Yes
Pub place/co/date	Yes	Yes	Called Imprint
Physical Descrip	Yes	Yes	Yes
Biblio Notes	Yes	Yes	Yes
Other Notes	Yes	Yes	No
LCSH Headings	2p L,Yes	L	L
ISBN	Yes	Yes	No
Location	Yes	Yes, bldg & flr	Yes, bldg, collec
Call Number	Looks like but not shelflist	L	Yes
Copies	No	Yes	No
Availability	No	Yes	Yes, due date if appl
Print/Download/Email	No-Frames	Yes	No
Other This Record	None	Lib/Search/Query Result#&Total, Relevance	Lib/Search/Query Result#&Total
Other Noted Elsewhere	Add Author/Title 2p. Link	L Add Author/Title	L Add Author inconsist L Series
MARC Available	Yes, complete, clear	Yes, complete, clear	Yes

Most catalogs available will search an accurate title effectively, although the information a patron has is often just the subtitle, so it is essential that this field be included in the search. None of the catalogs analyzed showed any problems in title searches. As noted, the Brandeis catalog even reported the full 245 field in each result. However, when the user wishes to search an author or name, many potential problems exist caused by choice of default search, search descriptions, or files or fields excluded in the search. It should be noted that the searcher might want items by a name, items about a name or both. Baring very clear search descriptions to the contrary, or a limit a user inputs, the default search should retrieve both. The following table presents the results when searching “Saint-Exupery” in each catalog. For the best search in each catalog, the menu position is in parentheses.

NAME/AUTHOR SEARCH IN DEFAULT AND OTHER MAIN PAGE SEARCHES

AUTOMATION PRODUCT AND UNIVERSITY LIBRARY OPAC DESCRIBED

RESULTS/ BY SEARCH TYPE	ExLibris Brandeis University	Endeavor Illinois State Univ	Carl.Solution U. of Maryland, CP
Default	Keyword, Any 37/37 100 %	Any Word, Any 100%?+1000’s irrel.	Keyword T. A. S. 0%
Boolean	n/a	62/62 100% (#4)	Only avail adv
Subject Browse	20/37 54%	34/62 50%, 20 headings, Linked	30/61 49%, 20 headings, Linked
Author Browse	22/37 59%	27/62 44%	n/a
Prolific Author Browse	n/a	26/62 42%	n/a
Call Number Browse	23/37 62%	32/62 50%	37/61 61%
Subjects	19/37 51%	n/a	0%
Name Author/Headings	n/a called author	n/a	0%
Name Author Browse	n/a	n/a	61/61 100% (#5)
Also Available	Beg.Title/Beg. Magazine	Beg.Title/Beg. Magazine	Title Br/Series Br

The user’s query phrase can make a big difference in the results gathered for a complex subject. Horn (2002) studied this problem using the single uncontrolled vocabulary word “garbage”. Testing several catalogs, degree of success varied widely. I chose to test a longer phrase. If the first word in the string does not match the first word in a subject heading, many catalogs will return a “no results” response, or simply place the user into the alphabetical listings. The second option is usually not immediately helpful, but may indicate to the user that more general terms should be tried. To demonstrate problems and capabilities of the three catalogs in this area I searched “metastasized breast cancer.” Results are found in the table below. Only applicable searches are included. See above table for all searches offered on initial search page.

SUBJECT SEARCHING IN NATURAL LANGUAGE ADJECTIVE FIRST

AUTOMATION PRODUCT AND UNIVERSITY LIBRARY OPAC DESCRIBED

# RESULTS BY SEARCH TYPE	ExLibris Brandeis Universtiy	Endeavor Illinois State U.	Carl.Solution U. of Maryland, CP
Default	0, tips with term permutations, has place of # items, link, not active	1721, only 1 indicated relevant	0
Boolean	n/a	with ANDs, 0	n/a
Subject Browse	0, tips as in Default	0, headings	0
Subject	metastasis-congress 1		0
Title Browse	0	0, hints, union	0

It seemed to me that most catalog users, especially with the Brandeis catalog, would understand the need to change their search to “breast cancer”. Therefore, I performed that search in all catalogs. The results are printed in the table below. “Breast cancer metastasized” is just as effective as “breast cancer” in the Illinois State University catalog with an Any Word, Anywhere search. In the other catalogs, however, attempting to retain this term at the end of the string, at least in the searches available on the initial page, sends the results back down to near 0 in any search.

SUBJECT SEARCHING CHANGE TO MORE GENERAL TERMS

AUTOMATION PRODUCT AND UNIVERSITY OPAC DESCRIBED

# RESULTS BY SEARCH TYPE	ExLibris Brandeis Univ	Endeavor Illinois State Univ.	Carl.Solution U. of Maryland, CP
Default	1721, overwhelming	1721, overwhelming	326
Boolean	n/a	188
Subject Browse	88 LCSH with # titles each, Link	Lots pages LCSH, MeSH, L, effective	Same list as below, but no # LCSH
Subject	Same as abovd		122 LCSH Linked, effective
Title Browse	33	31	62

COMPARISON OF THE THREE CATALOGS TESTED

One of the major problems faced by catalog designers is the need to make a catalog that an unsophisticated user can use successfully, without frustrating other searchers who are familiar with Boolean searches and advanced search forms. From my point of view, as someone who is headed for a public library position, automation products must certainly cater to the naïve searcher. Many of these users are aware that with a general search engine like Google.com they can use natural language to product good results, and with a retailer like Amazon.com they can have these results rich in content as well. These users expect their library catalogs to be just as simple.

In my testing of the three catalogs, I tried to approach them as a novice searcher would. I was most concerned with finding a catalog that looked simple, was easy to use, and would produce good results. In that regard, I did not use advanced searches, unless thrown into them by a default function, and I did not use Boolean terms unless the instructions were so minimal and clear and adjacent to the search box, that I thought only the most obtuse wouldn’t understand them. In the case of my subject searches, I used natural language, purposely choosing one that most probably would connect to controlled vocabulary, and one that wouldn’t. I wanted to see if these newer products would make any adjustments to assist the novice searcher, and in fact, there were some. I’m not totally convinced that they happen soon enough to capture the impatient searcher, looking for instant results, but the capabilities are definitely there. Two of the products, if adjusted in some important ways, might pass inspection. I would say that the third, while first attracting me with its simple design, proved a major disappointment. The Carl.Solution product used by the Library at the University of Maryland at College Park, and by the Union catalog to which it belongs, has made a large effort to keep the initial search page simple, at least visually. A little color, and a lot of white space make it visually appealing. The unsophisticated searcher is likely to focus only on the two simple boxes. One is already filled with the term “keyword”, the instructions above the other say “type in an author, title, or subject, and click search.” That sounds simple enough.

A problem faced by vendors is that they must not only design something effective with the “perpetual novices” described earlier, but satisfy the sophisticated searcher and the ones in-between as well. A sophisticated searcher, who knows how to type commands, might be satisfied with the advanced search in this catalog, which is for the most part, unguided. However, for one with mid-level expertise this might prove daunting, as would the “Help” area. Clicking on this button delivers the searcher to “How to Search”, a presentation which is very readable, but seems to scroll on forever. It is easy to miss the fact that there is actually a linked contents menu at the beginning, and a link back to it at intervals in the text. To me it seems that the vendor or librarian designer has made a questionable choice in placing the searcher in the midst of the text. At this point, it seems relevant to point out that vendors are making more options available to the purchasers of their products. It is really impossible for me as a tester to know whether a problem is solely the fault of the vendor, or the fault of the library system in choosing among options. For example, one of libraries on Illinet has chosen a color scheme for their initial search page that is to me unsettling enough to drive me away at first glance. The union catalog and all the other members I checked all used the same automation product, but had a very different color scheme, one of the best I saw. This is clearly a mistake on the part of the library, not the vendor. The vendor behind the library website will not even enter the minds of the vast majority of searchers. They will blame any problems they experience on the library or system.

Now it wouldn’t take most users too long to realize that that simple keyword box on the initial search page at the University of Maryland library website is actually a pull-down menu. This complicates things a great deal. As Yee and Shatford (1998) point out in their book, Improving Online Public Access Catalogs, normal users are not familiar enough with library systems to understand what files a particular search will cover, unless they are defined very clearly in words a novice can understand. The suggest having only a few options, and choosing the default in particular, very carefully. In the case of the University of Maryland not many users would understand the difference between an “Author/Name Headings” search, and an “Author/Name Browse”. Since with the former, even searching the exact authority record term will produce no results, I can only assume it is a disabled search. What then is it doing within the initial page pull-down menu?

Yee and Shatford might give this vendor some points for including “name” with “author” in the search term, but what they were suggesting was a search that proceeded through the authority file with any name (to locate the proper form) and on to the bibliographic record to find it in any field, presenting both the authority record(s) with attached files, and all bibliographic results to the searcher. As our class knows from our work with the Saint-Exupery shelflist, a searcher sometimes wants a results list that includes all works by an author and all works about them as well. The choice for default search in the University of Maryland catalog is entitled “keyword”, but it is far from the search Yee and Shatford envisioned. It searches very few fields. In a search for Saint-Exupery it produced no results at all, although the instructions on that very simple initial search page indicate than an author search is possible. You can actually find all Saint-Exupery records in one search using this catalog, and you can arrange them by call number, but you must use trial and error through at least two other searches in order to find the “Name/Author Browse” which accomplishes this. As I have indicated, the bulk of users do not last that long.

What excited me about the University of Maryland catalog in the first place was that despite its simple look, it certainly had more options than CONSULS, which is also a CARL product. The sorting choices looked interesting, and I experienced an interesting default into the advance search following a no-result search. This feature might be effective in encouraging the unsophisticated searcher to use this option. I was, for example, able to pull up two records in an otherwise unfruitful search for “metastasized breast cancer,” without simplifying the term. I also was successful in using the suggested added word to pull up only Saint-Exupery biographies. However, for the most part, in-depth testing of this catalog proved disappointing. Although it was possible to pull up all Saint-Exupery records in one search, it was not the default, but the fifth search on the menu. The default sort of the records, by date, proved helpful in verifying that there were no duplicate records, another problem experienced with the CONSULS catalog. The call number sort, however, proved disappointing, since clicking on the choice produced a change, but since the call number is not on the results pages, the process of creating a shelflist with this catalog would still entail the tedious process of clicking into every bibliographic record, and cutting and pasting. Also, none of the cataloging used in the three systems was any more likely than CONSULS to actually have Saint-Exupery materials within his one call number designation. It would be easier than with CONSULS, but not enough for me to rate the catalog very highly overall. I had far too many experiences of no, or only partial results.

The other two catalogs tested were on the whole, more satisfactory. However, if one looks at the choice of default search the catalog of the Illinois State University System and the union catalog to which it belongs, this Endeavor product also comes up lacking. I suppose some might feel that this Any Word, Anywhere search is superior. In all of my searches it did produce results. The problem is, it produced way too many results, over 1000 in every case, and over 8000 when searching “solar energy.” The relevance sorting did appear to produce seemingly good results at the top of most lists. However, in the case of the “metastasized breast cancer” search the results in the initial pages of 1721 results appeared to have nothing to do with the subject at all. Novice users give up in the face of no results, but they also give up in the face of too many. Even though I know how to limit results, I prefer to not be faced with so many to begin with.

The University of Maryland catalog also still had too many searches on the initial page to avoid frightening away a novice, and since there is an “other search” button, the total number offered from this initial page is 9. Admittedly, academic libraries may have more sophisticated users overall than public ones. Though once one moves beyond the default search, I like may of the searches offered, I don’t think they all need to be on the initial page. Adding more to the “other searches” page would be a first step. I would actually make the Boolean search, #2 on the list, the default. I think the simple instruction provided “use and, or, or not” is simple enough so that anyone could figure it out with a minimum of trial and error. There are also other tips right on the initial screen. The user doesn’t have to click on “Help” to find them. All the search types are covered in no more than 3 inches of screen each, so even to the novice user, it might not look too hard. It’s certainly convenient. The Boolean search did pull up all Saint-Exupery records. This catalog doesn’t offer a call number sort, although it does offer a call number browse. It was successful on all the other searches except the “metastasized breast cancer” one. Here it suggested using the union catalog or changing search terms, although these were placed in an area where the eye is used to seeing text, so they might not be read. Once I did broaden my terms, the Subject Browse was extremely helpful in connecting me with the very powerful sorting tool of LCSH. Given the huge list of headings provided, as a user I would be fairly confident I could find some relevant material without looking further into most of the headings. Yee and Stratford (1998) point out that the use of keyword searches as the default setting effectively bypasses the sorting for most users who will not know to look for the search that accesses that file. They also note that most libraries only include within their authority files the headings actually used in their collection. I don’t think this is sufficient however, as evidenced by the sparse headings offered by the University of Maryland catalog, far to few to provide help in my breast cancer search. I appreciate the problem of putting all LCSH on every catalog site, but I don’t understand why each catalog can’t provide a prominent link to them to help users discover the proper terminology to use in a complicated search

Despite the number of searches on the initial page, it appears fairly simple, with pleasing neutral colors. It could appear more simple by using the University of Maryland’s technique of burying the multiple searches in a pull-down menu, but as I said, I would tend to choose just 3 instead, moving the others to another page. Beyond the searches, the options on the initial search page are minimal, but once you reach a results screen, the power of this catalog becomes apparent. There are multiple ways to limit your search results both right on the screen and through a link. You can also sort your results in four different ways. One of my pet peeves is having only 10 results on the screen, and having no option to change it. This catalog gets high marks from me for having the default be 20, and allowing up to 50 by a change of preference. For extended searching, the navigation on this site is great. The search term will stay in the box, although a clear button will erase it if you wish. There is also a search box right on the results page, which isn’t always the case. Each result also contains the full title, the first author, the date, the format, the call number, the location, and the availability. This is not adjustable as in the Brandeis catalog, but it seems sufficient. Only the title and the availability are selectable however.

The bibliographic records on this site are very nice with many of the fields linkable in one step. My only complaint is the brief/full/MARC choice buttons which are in a nice place, but are small dark and hard to read. Once you find them, the MARC records are beautiful and complete. The Brandeis catalog has nice bibliographic records, including MARC as well, but though some fields are selectable, it is a two-click process that is inefficient. No one had the help for students to email, download, and print that were available through the Illinois State catalog, and this is a feature that particularly students really want. The Brandeis catalog did have a button to create a no-frame form.

In some ways, the Brandeis catalog is my favorite in design, but I could never choose it overall, because at least from across the country, though I could get onto the homepage, I could often not access the catalog. There were also massive problems with timeouts in an extended search, as there were with the other two catalogs as well. That is one area where I would definitely prefer CONSULS, although that may be because I am a registered student. The University of Washington catalog does not through me off as much either. The primary attraction of the Brandeis catalog is its functional success. The default search pulled all Saint-Exupery records. The default settings on result presentation are also very complete, and even more amazing it is customizable as well. The sight appears very navigable with two rows of text choices at the tope of every page. It is the only sight to have a “basket” which Horn (2002) reports that patrons want. They want the library to be more like a commercial site with a basket or cart. I have gotten used to the marked record procedures, and calling it a “basket” is no improvement as far as I am concerned. The site also loses my regard in the fact it has no search box on the results page, so you have to back out to the initial page to search again, and your term is not saved.

Among the products I tested or looked at I would choose the Endeavor system, but only if the default search could be changed. It is probably still necessary to look to the future in hopes of improvements to find a catalog that performs well on all scores, and even then, our customers will expect changing content and features. It is becoming the norm all over the Web. Developments I would like to see are increased content. This is available on the King County Dynix powered IPAC, though of course only with new records, which is a problem that extends into the completeness of bibliographic records online as well. I would also look for that link to LCSH headings, and integration of collection, periodical articles, databases, and outside library links as well, with a dynamic navigation between them. The Gale database is already adding MARC subject headings to its records and will appear soon in the GEAC catalog described in the Puacz (2002) article. It is an exciting time, but it is essential the library keep up with the competition in the public sector.

BIBLIOGRAPHY

Conrad, Deby and Laurie Lessner (2002). Setting a course for the 21st century. Computers in Libraries, Oct. 2002, p. 23-25.

Horn, Marguerite M. (1998) “Garbage” in, “refuse and refuse disposal” out. Library Resources and Technical Services, 46:3, Jully 2002, p, 97-101.

Puacz, Jeanne Holba (2002) Catching and keeping e-patrons. Computer Libraries, 22:1, January 2002, p. 13, 4 pages.

Yee, Martha M. and Sara Shatford Layne (1998) Improving online public access catalogs. Chicago: American Library Association, 1998, 220 pages.

Cataloging Issues
Home

;