NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Institute of Medicine (US) Committee on Internet Access to the National Library of Medicine's Toxicology and Environmental Health Databases; Liverman CT, Fulco CE, Kipen HM, editors. Internet Access to the National Library of Medicine's Toxicology and Environmental Health Databases. Washington (DC): National Academies Press (US); 1998.

Cover of Internet Access to the National Library of Medicine's Toxicology and Environmental Health Databases

Internet Access to the National Library of Medicine's Toxicology and Environmental Health Databases.

Show details

2Assessment of the Current TOXNET Web Site

The 1997 IOM report on the utility of NLM's toxicology and environmental health databases identified access and navigation as key barriers. At that time, the databases were primarily available on a cost-per-search basis (costs depended on the amount of online time and the type of activity) that required registration for an NLM password and provisions for billing. Additionally, the primary method for searching was through a command line interface, which required knowledge of a complex search language and syntax (e.g., field names, print commands). To those who frequently use that interface to search the databases (this option is still available), the command line interface can be a powerful tool for search retrieval, but to infrequent or novice searchers, it is largely impenetrable. While pre-Internet Grateful Med offered an additional search option for most of the TOXNET databases—one with a more user-friendly interface—it required purchase and loading of the software.

The move to a free, Web-based interface represents a major leap forward in providing access to the TOXNET databases. The committee believes that this set of databases can provide valuable environmental health information to a wide range of users. The committee applauds NLM and particularly the staff of the Specialized Information Services Division (SIS) for their commitment to increasing accessibility to the TOXNET databases.

The committee understood, as they began their deliberations, that the Web site was a work in progress, would continue to undergo changes during the course of the study, and would continue to evolve once launched on the Internet for public access. For the majority of the time that the committee was testing the Web site and seeking input, the site only provided access to five of the factual databases on TOXNET (HSDB, CCRIS, IRIS, RTECS, and GENE-TOX). Search screens have been added more recently for the bibliographic databases—DART, EMIC, EMICBACK, and ETIC—as well as for the TRI databases. As a result, the committee's assessment is primarily focused on the factual database search screen. However, a number of the recommendations may be generalized to the entire TOXNET Web site.

This chapter summarizes the search experiences of the committee and of the individuals who provided input to the committee (Chapter 1). Most individuals searching the TOXNET Web site experienced similar problems. Those problems are highlighted in this chapter's assessment of the current search interface and a number of refinements are noted.

Selecting a Database

One of the issues associated with the TOXNET system, no matter what the interface, has been the fact that it contains numerous specialized databases (e.g., DART and ETICBACK that focus on the teratology literature), some of which are familiar only to a focused user community. Therefore, the novice or infrequent user may have difficulty in determining which database to select to begin searching.

In most cases, searchers chose the default database, HSDB, as their initial selection. The committee agrees that HSDB should be the default database because it is has a wider range of information than the other factual TOXNET databases. Additional mechanisms are needed for informing the searcher about the content of each of the databases. While, the current help screens do an excellent job of explaining each database and providing the searcher with information on its purpose and structure, few searchers in the committee's experience took the time to read the help information (see later discussion). Options for improving assistance in database selection include:

  • an opening screen with information on each database and guidance in selecting a database;
  • a hypertext link to information about each database, integrating the help information into the search screen; or
  • step-by-step screens that walk the user through the database selection process (see Chapter 3).

Inputting the Search Strategy

The current search form for the factual databases provides one box to enter chemical/other names (e.g., chemical name fragment or Registry Number) and one box to enter subject terms. The form then provides a list of LOOK FOR options for combining the subject terms and another list of options for displaying results.

There were mixed reactions to this search screen setup. Some searchers found it too complex, preferring one box to enter all search terms (e.g., Yahoo or PubMed). Other searchers wanted to be able to perform complex Boolean searching and would have preferred multiple boxes for entering chemical names and search terms (e.g., Internet Grateful Med). Thus, the committee concluded that the current search screen is too complex for simple searches but does not offer the specificity and complexity needed for advanced searching. This disparity indicates the need for several levels of search interfaces, from beginner to advanced, to meet each individual's search needs (see Chapter 3).

After entering chemical and/or related subject information, the next decision point in entering the search strategy is to choose how to combine the search terms. The current search interface has the following list of LOOK FOR options:

All subject terms within record (AND)

Any subject terms within record (OR)

Subject terms as phrase (NEAR)

Subject terms in same occurrence (SAME).

Searchers were unsure about what those choices meant. The options are not listed from broadest retrieval (OR) to narrowest retrieval (NEAR). Even individuals familiar with Boolean searching were not clear about the NEAR and SAME options and related concepts. Searchers would have preferred to enter phrases encased by quotations rather than use the term NEAR, which searches within six words. Additionally, because of the lack of understanding of the basic structure of the TOXNET database records (discussed below), searchers were not sure to what extent choosing the SAME option (same occurrence) would narrow their search. As in selecting a database, the help screens did offer examples and additional information. Those clearer explanations should be incorporated into the search screen either as hypertext links or as additional text on the search screen itself. The committee makes recommendations in Chapter 3 for varied levels of search screens that would assist in addressing and clarifying these issues.

Displaying the Search Results

One of the most difficult problems for searchers was in understanding the search results—both how to display the information on the screen and how to interpret the displayed information. Most users (e.g., health professionals and students) are more familiar with searching bibliographic databases, particularly MEDLINE, where searches on a chemical retrieve numerous citations. They were not familiar with the term ''record'' or with the structure of the TOXNET factual database records. Each record generally contains all the information in the database on the chemical organized into numerous fields. Searchers found it helpful to have the search terms highlighted in the retrieved text. However, reading the search results on the screen required a great deal of horizontal scrolling on some computer monitors.

Other problems encountered relate to the five display or view options: chemical listing, exact term match (the default option), full record(s), customized display, or selectable list (all results). Searchers did not understand the differences among these options. The following three examples illustrate common problems.

For searchers who entered the term "vinyl chloride" in the chemical name field and kept the default display choice of "Exact term matches," the retrieval displayed the message "Your search found 1 record" with two lines of text (name and registry number) and the message "End of Results'' (Box 2.1). This was disconcerting to many searchers who expected numerous "hits'' or records. Additionally, many were unclear about why they only received two lines of text. Novice searchers felt that the two lines of text retrieved were the extent of the information on vinyl chloride in that database (largely because of the message, "end of results"). They then proceeded to choose another database in the hopes of retrieving additional information.

Box Icon


BOX 2.1. Your search found 1 record in Database: HSDB Query: ((NAME) VINYL CHLORIDE)

A second problem presented itself to searchers who chose the display option, "Full record" and retrieved the first page of the record. They realized there were multiple pages that followed but would have preferred to know the extent of the record (how many pages or bytes) without paging through the entire record to determine its length. Further, most searchers were unfamiliar with the structure of the record. They were not sure, for example, where human toxicity information would be found or if the database contained that information. Searchers who confronted this problem would have preferred to have an outline of the record or a hierarchical display of the major fields with hypertext links to sections of the search results.

The third, display-related issue came up with searchers who were sent to the middle of the record. These searchers frequently searched on both the chemical name and a subject term (e.g., vinyl chloride and liver cancer) and chose the display option "Exact term match." In HSDB, the retrieved information on that search begins in the vinyl chloride record in the field on storage conditions, followed by a field for emergency medical treatment information, and then information on clinical effects. The searchers found this information useful in answering the search question, but they were not sure about the context of the information or what other information was available in the record.

Searchers were pleased to find or be told about the "customized display" option; they felt that this option solved many of the problems described above because it provided an overview of the record content and presented them with useful options to narrow their search (e.g., human health effects, animal toxicity studies, emergency medical treatment). Changing the default display option to "customized display" or presenting an outline or hierarchy of the database record would be useful in familiarizing searchers with the content and structure of the database. "Customized display" is a valuable option that should be made more visible to the searcher as most searchers did not understand all that this option has to offer.

Modifying the Search

Once the initial search was performed and the searcher had assessed the results, there was frequently a need to modify the search. It was clear to searchers that they could either return to the search screen by way of the BACK button on the Web browser or by selecting the MODIFY SEARCH button in the left frame.

Several searchers wanted to work with only the subset of records they had previously retrieved to further narrow the search results. Currently, the search criteria can be modified, however, each search is run anew on the entire database. This problem is in part a function of the Internet and the difficulties involved in storing sets for each search that can then be combined or manipulated (Chu, 1998). Further refinements are needed to allow searchers to have increased specificity in modifying their search (e.g., PubMed's Advanced Search) and to work with stored sets of search results.

Printing or Downloading the Search Results

The final step of the search process involved saving the search results in hard copy or online. Printing the search results presented problems because each page had to be viewed and printed individually. Searchers would have preferred the option to print specific fields or to print the entire search result. This is a problem of which the NLM staff is aware. However, it is difficult to address because it is largely a function of the database file structure. Having the capacity to print all or a specified part of the record would be optimal.

Searchers utilized the e-mail option once they found that it was available. However, because it is currently listed under the redisplay options, it was not readily apparent. It would be preferable to have more obvious buttons or links at the top and bottom of the screen for displaying, retrieving via e-mail, or printing the search results. Additionally, a download option (save to disk) was requested by many searchers.

Other Issues

One of the advantages of the Web is that it facilitates feedback. Currently the TOXNET Web site offers searchers the opportunity to send an e-mail to TOXNET User Support to discuss problems, request assistance, or provide comments or suggestions. This feedback is important and can be supplemented on occasion by seeking searcher input on specific features of the TOXNET Web site.

The use of frames, a topic of ongoing discussion (Siegel, 1997), was an issue with a number of the searchers. The committee felt that frames did not add to the functionality of the TOXNET Web site. The left frame, containing options for modifying the search, new search, online help, and searches on other databases, currently remains relatively static once the search has begun. Depending on the computer monitor, the size of the right frame was limited and required constant horizontal scrolling to read the search results on screen.

If a design decision to retain the frames is made, then they should be used to enhance the site. Examples of possible frame uses include providing context-sensitive help (Chapter 3), listing the major sections of the database record (e.g., toxicity and biomedical effects, pharmacology), and providing links to jump to specific parts of the record. However, in the committee's view, the site would be better served by removing the frames and adding buttons and links as needed.

Other factors addressed by the searchers included positive experiences with the response time of the search and a few problems with connections to the Web site. The move of the TOXNET system in May, 1998 to a new hardware platform appears to the committee to have gone quite smoothly, and the new hardware system seems to have speeded up the search processing.

Copyright © 1998, National Academy of Sciences.
Bookshelf ID: NBK45167


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (213K)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...