Fulltext Databases: A Promising Utility for Document Delivery
Reference Librarian, Electronic Services
Heindel Library, Penn State Harrisburg
Middletown, PA 17057
Fulltext databases are charming as well as challenging to libraries. Today, there are many types of information available in fulltext databases, such as books, periodicals, newspapers, reference materials, as well as electronic publications. This paper focuses on the electronic versions of newspapers and periodicals, either in ASCII format or image format. I will discuss how fulltext databases are used as part of our library collection, and issues to consider.
I. Fulltext databases: Types and Usages
The library I am working for is a branch library in the Penn State University Libraries System. It is located 100 miles away from the main campus. It serves about 150 faculty members, 3,500 upper division and graduate students, and southern central Pennsylvania community. (For more information about the college, please check our home page at http://www.hbg.psu.edu/Hbg/pshoverv.html.) To serve this user body, we have included more and more fulltext databases into our digital library collection in recent years.
Lexis/Nexis is available with online subscription. We access Lexis/Nexis databases via the Internet connection. Under Lexis/Nexis's educational license agreement, library patrons (restricted to college faculty, staff, and students only) have unlimited access to a wide range of Lexis/Nexis databases in the library. An annual publication is provided by Lexis/Nexis with indication of three levels of coverage: complete fulltext, selected fulltext, and abstracts only. All covered articles are provided in ASCII format. The search functions include fulltext search, single publication search, single date search, etc. With the [PrintScreen] command, patrons can print or download a fulltext article screen by screen.
2. Dow Jones News/Retrieval
Dow Jones News/Retrieval (DJN/R) is also available via online subscription at the educational rate. It is provided for in-library use with linkage to DJN/R via the Internet. Under our license agreement, we don't have access to DJN/R's text databases. But we still have access to more than 600 publications in ASCII format through DOWQUEST section, including business and financial publications such as Wall Street Journal (WSJ). Print and download functions are both present. It is an enhancement to Lexis/Nexis because WSJ is not available in full text in Lexis/Nexis. There are drawbacks in DJN/R, however. Firstly, the list of coverage has been dropped from their recent users guide. Secondly, search interface provides no mechanism for known-item retrieval. Lastly, it only covers articles in the last six months. (Note: after submission of this article, the author found that DJN/R had removed DJREQUEST interface.)
3. Computer Select
Computer Select is a CD-ROM product from Information Access Company with a roll-over 12 month coverage of 113 computer journals, industrial newsletters, etc. Selected fulltext articles from about twenty-five publications are provided in the CD-ROM in ASCII format. The database has many search keys including single publication title and single date. Both print and download functions are available. The CD-ROM database is available for in-house library use only. We did not receive any updated coverage list since the initial subscription.
4. UMI Business Periodicals Ondisc
The Business Periodicals Ondisc (BPO), the fulltext version of ABI/Inform, is the only database in the library which provides a true cover-to-cover image coverage of more than 400 periodical titles. Some titles go back as early as 1987. The BPO is installed in the library, whose CD-ROM collection amounts to 500 by now. Two types of search engines are available: 1) conventional UMI ProQuest interface, and 2) a periodical directory. Using the periodical directory is pretty easy. User may trace an article by starting with periodical title, issue number, and then article title in a table-of-contents. With a laser printer, the quality of printouts is similar to a black-and-white photo copy. However, there is no download function for articles in image format. In our library, we have incorporated the BPO coverage into the library periodicals holding list and updated them annually with information provided by UMI.
5. Periodical Abstracts with Fulltext
This service, using UMI tape-load data, is available through Penn State university-wide Libraries Information Access System (LIAS) in ASCII format. Some fulltext coverage began in 1992, and many titles started after 1994. Only major articles are covered for each issue of publications. Through the LIAS interface, only the abstract parts of documents are searchable. It is possible to get a list of covered articles in a single issue of publication. Print, download, and email functions are provided. Library patrons can access this database from any Penn State libraries or from a remote location. A coverage list is provided.
Databases or services listed above have other coverage beyond fulltext newspaper and periodicals. They are not included in the discussion of this paper.
Fulltext databases have many unique advantages. Compared with the conventional library collections, fulltext databases are easier to store and search. Moreover, it is easier to deliver documents originally published in newspaper and periodicals with fulltext databases. They take much less shelf space to store. They come with all kinds of search engines. Contents can be printed or displayed as one wishes and are remotely accessible if permitted by the license and technology. But selecting and using fulltext databases can be difficult or confusing. Here are some major issues we should deal with regarding using fulltext databases.
1. Image vs Text
It should be clear from the start that two basic types of fulltext databases -- image-based or text-based -- serve different needs. Image format preserves the original document in verbatim, including tables and diagrams. (UMI's BPO can't handle color or gray scale images well, though.) There are many valuable information embedded in the layout of the original publication. When transferred to ASCII codes, many articles lose accompanying information, such as tables, diagrams and pictures. But text format enables users to print or download a document more conveniently with less technical requirements. It is especially true when one accesses the document remotely.
Our library patrons have a strong preference towards BPO over the printed publications, for it is more convenient to use than printed stacks in many ways. For text-based databases, on the other hand, Lexis/Nexis newspaper collection is an extreme success. Although articles in Lexis/Nexis have all the disadvantages of being in text format, they are preferred to microform copy. In Lexis/Nexis, patrons can easily search a known newspaper article. In addition, it is searchable by date in a single newspaper so that a patron can "browse/scan" newspapers more easily.
Today, fulltext databases in text format are more readily available than image-based ones. But popularity does not determine which format is superior to the other. In many cases, patrons are in the best position to decide if their needs can be met by using an image format or text format. Having access to a publication in text format does not means that every patrons can take advantage of this format. However, if librarians show patrons how to access to those fulltext databases, they will certainly offer more resources than otherwise available in the library.
2. Full Coverage
Not all fulltext databases deliver the promise of fulltext coverage. The completeness of coverage is a very important factor to consider including fulltext databases into your library collection, not even mention replacing some titles in our collection. Ideally, we should look for those with truly cover-to-cover coverage. Again, a good example is UMI's BPO. With BPO, you can access any content, even by page numbers, in a single issue.
In reality, however, we may or have to settle down with some databases that have less coverage. The real problem with fulltext databases is that many times we are left untold as to what articles are selected and what articles are not. Another problem related to the incomplete coverage is that the database users, either patrons or librarians, are forced to check it very closely before they can be sure that they do not have the article buried somewhere in the fulltext collection. It is a time-consuming task. Lexis/Nexis's coverage list is a useful tool for this situation, for it has indications for different coverage levels. UMI's coverage list, accessible through the online fulltext periodicals directory, is also very handy to identify if an issue is completely missed.
3. Known-Item Retrieval
When examining these fulltext databases, one important feature that should not be neglected is the "known-item retrieval" capability. A fulltext database should allow searching and retrieval of a known-item, that is, a specific article that a patron knows its title and/or author.
Ideally, the searching should be quick and straightforward, and the result matches the intended hit as closely as possible. A search result shouldn't need scrolling more than two screens. In another word, the search engine should give patrons explicit results whether the articles exist or not in the database as quick as possible.
The DOWQUEST in DJN/R is a negative example. Actually, the DOWQUEST's search engine is relative easy for subject search. It serves its purpose of finding "relevant articles" quickly and easily. But it is very difficult to search an article based on some known citation information which are non-subject related. The search inquiry is always treated as a statement in the "nature language" and the results are usually four screen long. Chances are that the article needed is not shown up on any of those screens.
It should be noted that fulltext databases tend to cover popular publications, many of which may have been in your library's subscription already. It is possible to see a heavy usages of these titles in fulltext databases. But again, fulltext databases, no matter how complete they are and how conveniently they can be accessed, are not the same as the printed counterparts. The digitized version normally comes later than the printed version, except some newspapers in Lexis/Nexis. The digitized version can't be viewed without a computer. The printout from a fulltext database may not as readable as the original printed version. It is always hard to browse a fulltext database. Thus, a library can't afford to cut subscriptions to those popular titles and simply live with fulltext databases. It should be kept in mind that fulltext databases may serve some needs but not all needs. For archival purposes, however, a library may choose not to bind, store or order the microformed copy of some newspapers or industry newsletters.
On the other hand, it should be aware that there is always some duplications among several fulltext databases. Once a publisher makes its publication available in digitized format, it may lease the contents to more than one database vendor. For example, PC Magazine, Fortune, Business Weeks, etc., can be easily found in several fulltext databases mentioned above with the same contents. When evaluating those fulltext databases, we should discount the value of these duplicated publications.
By introducing fulltext databases into library collection, we are gradually moving away from a collection-centered to a services-centered approach. Thus, the quality of services provided by database vendors are particular important to a successful use of fulltext databases.
For instance, can the vendor provide the fulltext databases in a long run? We know that the fulltext databases may partially replace some of archival collection. The problem is that we are not sure if the vendor can last as long as we are still in need of those databases. Even databases are still available, some publication titles may be dropped along the way. Far East Economic Reviews was considered as one of important journals we could get through BPO. But the title was dropped out of BPO in 1995.
Does the vendor have a reasonable license arrangement and price? For example, can we provide remote access to the database? Can we use the databases for a college wide need? Lexis/Nexis's educational license clearly states that the products can only be used for a curriculum related purpose. Using Lexis/Nexis materials for research, reference, and college administration is prohibited under our current contract.
Does the vendor provide a coverage list that reflects the contents of the database? Do they update the coverage list regularly and accurately? We have discussed this issue above.
Fulltext databases are a promising phenomenon. They bring in a new service dimension to libraries. From the viewpoint of the collection development, they are strengthening our collection while competing for limited resources. Based on my observation, it is clear that fulltext databases can not completely replace our conventional collection. However, they can be used to replace some piles of back issues of newspapers and newsletters. Beyond that, we still need to ask for more, better, and truly complete fulltext databases.
Copyright © 1996 Huijie Chen.
Submitted to CALA E-J on May 1, 1996.