Eric T. Rumsey, Librarian, Hardin Library for the Health
Sciences, University of IowaIowa City
Adapted from a talk given at the Collection Development, Public Services, and Technical Services Sections, Medical Library Association annual meeting, Chicago, May, 1999 [Abstract]
Hardin Meta Directory - hardinmd.lib.uiowa.edu/
Please send comments to firstname.lastname@example.org
In looking for information on the web, there are two basic approaches - Search engines and directory sites. Search engines work well when you're looking for a specific bit of information. But they don't work well for broad searches - In Alta Vista, for example, searching for "heart" gives 6 million links and searching for "cancer" gives 2 million links. For broad subject searching, directories work much better.
Directory sites are created by individuals, often originating from personal bookmark lists, which is in fact the origin of Yahoo, the prototypical directory site. Directory sites make it relatively easy to browse the resources in a subject, to see at a glance what sorts of resources are available in the subject.
It's interesting to note the parallel between these two types of web searching and the process of searching for books in a library using traditional library tools. Search engine searching can be seen as analagous to doing a keyword search in an OPAC, while directory searching is analagous to browsing the shelf, utilizing the library classification system. The Hardin Meta Directory is a tool for finding the best directory sites in health and medicine. As I have found in working on the meta directory, there is a remarkable number of good directory sites in health/medicine developed and maintained by librarians. I would like to suggest that one reason for this may be the long experience of librarians in providing browsing access to their collections by classification number.
A bit of background about the origin of the Hardin Meta Directory will help to describe its nature. It was born in late 1995 when I was looking for the best directory sites, or lists of sources on the web, in several areas of clinical medicine. At that time I thought that Yahoo was the best source for directory pages (as many people still do!). Going to Yahoo, I found that, indeed, it did have seemingly long lists in all areas of clinical medicine. As I did more digging, however, I found that there were actually directory sites that had much more comprehensive lists of links in clinical medicine, virtually all of which were done by people who specialize in health/medicine. In most cases these sites were relatively little known compared to Yahoo. Having found these lists and having added them to my own personal bookmark list, I thought, why not make these lists available to other people? And, that, in a nutshell, is the origin of the Hardin Meta Directory.
The original purpose of the Hardin Meta Directory - to lead the user to the most comprehensive directory lists - has developed a bit further as it has evolved. As I worked on it, I began to look for ways to evaluate the quality of directory sites, particularly the quality of their upkeep and maintenance. I began to realize that it's all too easy to create a list of bookmarks for a subject - Anyone can spend a few hours with a search engine, and make a tidy list to load on the web, and call it a directory site. But if the list is left unattended for six months or so, many of the links will go bad. The real work of doing a directory site is, not creating it, but keeping it maintained.
So I looked for ways to evaluate the level of maintenance of the directory sites that I was listing. At first, I used the date of last modification of the site, either taken from the page itself, or from the browser's Document Information menu. The problem with this is that this information is not always available. So I began to check connection rate. I started measuring connection rate in early 1997, when the meta directory was about a year old. I measure connection rate by using an automated link checker (Big Brother on the Mac) and by random sampling of links.
I was encouraged to find out, when I first began to check connection rate, that it correlated well with the modification date, which I had been checking previously. Sites with a good connection rate usually had a relatively recent modification date; sites with an old modification date almost invariably had a poor connection rate. So, gaining confidence in my measurement of connection rate, I began to make it known in the site documentation and on subject listservs that I was using connection rate as a criterion for placement on meta directory pages. At first I was rather apprehensive about this, thinking that list-keepers would question my checking of their sites. Surprisingly, however, there was little questioning. The few people who did question it were quickly turned around when I explained how I did link checking - In several cases people thanked me for making them aware of the problems with their site, and did a good job of cleaning up their bad links. The general lack of questioning, however, makes me think that most list-keepers are quite aware of the level of maintenance of their sites. If they have a poor connection rate, they know it!
After about a year of measuring connection rate I felt confident enough in my technique that I created the Hardin MD Clean Bill of Health <hardinmd.lib.uiowa.edu/cbh.html> award for sites with connection rates of 93% and over. This has been very successful, giving the site much good publicity.
Connection rate is certainly important in itself, but it also has importance beyond simply telling the proportion of links at a site that connect successfully; it also serves as an excellent general indicator of the level of care and attention that's given to a site. If a site has a good connection rate, it means that someone cares enough about the site to keep it maintained, not only making sure links are connecting, but probably also keeping it updated by continuing to add new links. So, in a sense, the Clean Bill of Health award actually does serve as a measure of the general health, or vitality, of a site. <hardinmd.lib.uiowa.edu/best.html>
In order to discuss features of the Hardin Meta Directory, I'll mostly use the cardiology page <www.lib.uiowa.edu/hardin/md/cardio.html> as an example, with brief mention of the rheumatology page <www.lib.uiowa.edu/hardin/md/rheum.html>. If you're reading this narrative with a browser, I'd suggest opening the two example pages in a separate window. If you're reading a paper copy, please print a copy of the example pages to follow the description.
The cardiology page is fairly typical in size, and in the types of sites it includes - It has representatives from many of the large "umbrella sites" in meta directory that have several medical/health subjects, including Medmark, Karolinska Institute, Mining Company, MedWebPlus, Yahoo, healthfinder, NOAH, SciCentral, and MedlinePlus. Several of these are from a library environment, as are the representatives clearly identified as such from the University of Washington and Yale University. It also includes independent sites that concentrate on cardiology. There are sites done by physicians and nurses, as well as librarians.
Particularly noteworthy features on the cardiology page -
-The indication of list size (large, medium, small) is an especially valuable feature in looking for comprehensive directory sites.
-The ranking of sites on the page is based on a combination of the size (number of links) and the connection rate.
-The first four links are to non-US sites - The web truly is a WORLD WIDE web - Surprisingly, many of the best directory sites in the meta directory are from the non-US world.
-Nine of the first ten links are from umbrella sites - The cardiology page was formerly dominated by independent sites that have fallen by the wayside in the last couple of years.
-Links are to the specific appropriate part of the page within the larger site. In the Martindale site, for example, this is very helpful, because it's often difficult to navigate to the relevant part of its long pages.
-9 of the 21 sites listed are from libraries or librarians.
One of the strengths of the Hardin Meta Directory is that it includes links to directory sites done by people with a wide variety of backgrounds. Many excellent sites that are included are done by people who are not practicing health professionals, including many by "patients." On the web, there tends to be a blurring of the separation between health professional and patient. A good example of this is on the rheumatology page - Two good lists on that page (Arthritis at About.com, Awesome arthritis links) are done by people who are in the category usually called "patients" - Close inspection, however, shows that they are both former health professionals (med lab tech, nurse) who have been forced to stop work in their professions because of disabilities, and are spending their time working on (high-quality!) web sites. The blurring between patients and health professionals is also illustrated on the cardiology page by the last link, from JAMA NetSight. Its title - "The Cardiology Beat: An Internet Education for Patients and Health Professionals" - shows that the AMA realizes that the same information that's useful for patients can also be useful for health professionals.
The Hardin Meta Directory is very much a library-oriented source - It has deep roots in the library world. As I've mentioned earlier, and as is clear from the Cardiology page, librarians have done a tremendous job in creating medical/health web directory sites. In preparing for this talk, I've done some thinking about just what it is about traditional library work that seems to serve as good preparation for the development of web directories. Because traditional library work is so different from web work, it's not easy to make direct comparisons. But, as I mentioned in the introduction, I think the best connection I can see is that the long experience of librarians in classifying books to make them accessible for users to browse has given us a good set of skills for organizing web directories. Additionally, for us, as medical librarians, I think the experience with the superlative MeSH system has given us even a further advantage.
As I've been doing since before developing the meta directory, I keep close track of medical/health library web sites, and in particular their "web directory" pages. Several libraries are represented with some of their directory pages. But it's not only sites like Yale and Washington that have their own subject directory pages that I find valuable - Librarians seem to have a good "nose" for quality directory sites, so I often get leads from library pages to other good directories.
There has been much discussion in the library world about whether librarians are in a position to "catalog the Internet." Many have expressed the opinion that we would do a better job at this than sites like Yahoo. My view is that we in the library world simply don't have the person-power to organize the net by ourselves. So what I've done is to include high-quality directory sites from wherever I find them. Because librarians seem to be well-suited to creating them, a high proportion of these are from the library world. But many are not. What I look for with the meta directory is a sort of "list-keeping instinct" - Librarians certainly have this quality in great measure, but the list-keeping instinct also burns strongly in many non-librarians around the web. So with the Hardin Meta Directory I'm able to combine the already well-developed list keeping talents of librarians with the nascent talents of folks around the web, nurturing them along as I find them.
In thinking about the "list-keeping instinct" in librarians, I think that, like other instincts, it can at times burn TOO strongly! - Looking around medical/health library sites, I'm often struck by how much duplication of effort arises from many different libraries producing their own lists of web sources in health/medicine. With the large number of excellent lists that already exist, it's really no longer necessary for each library to create their own. My suggestion for library list-keeping is that individual libraries would do well to make generally short lists for specific subjects, with strong emphasis on sites that are especially useful for their own, local users. There should be a clear statement accompanying the list, making it clear that what's being presented is not intended to be a comprehensive list, and links to truly comprehensive lists should be shown prominently. A good example of how this can be done is the SciCentral cardiology site. On this, there are a handful of comprehensive lists, including the Hardin Meta Directory, at the top of the page, then several specific links that are deemed to be of particular interest to SciCentral users.
So, in conclusion, I'd like to ask you all to take a careful look at the Hardin Meta Directory, and to consider making as many links to us as you want!
The Hardin Meta Directory of Internet Health Sources (Hardin MD) is a prime "starting point" for finding Web medical information. Created and maintained at the University of Iowa's Hardin Library, Hardin MD provides access to the best resource sites in 37 medical disciplines; its popularity is shown by the thousands of links to it from around the Web, and by reports on it in the medical/health sciences press and the lay press. In some ways, the work of the Hardin MD is very much like the traditional library world of reference, collection development and cataloging, the only difference being that the work of our analytical efforts are on Web resources instead of in books. The Web enables us to communicate the results of our work to the world beyond the doors of the library, and to work closely with people throughout the (library and non-library) world. The paper will discuss how this communication and collaboration has been achieved. A particular focus of the paper will be the use of link-checking as a quality control device for Web resources.