Techniques: Off the Shelf

Tired of searching the Web and hauling in junk? Here's an inexpensive search tool that gives you what you want

To professional researchers who are accustomed to using on-line information services where articles are indexed, abstracted, tagged, and structured to promote well-targeted retrieval, the World Wide Web is just a joke.

Although they pay dearly for the efficiency of on-line retrieval services, researchers consider that expense an extremely attractive alternative to wading through oceans of false Web hits. Web searches seem forever doomed to collect trash. After all, Web-site designers go out of their way to make sure search-engine users will stumble onto their sites. Even a very specific search can yield thousands of irrelevant hits and only a slim hope of any gold nuggets.

WebSleuth goes a long way toward solving that problem. The software simultaneously passes your very general search to such standard Internet search engines as AltaVista, Excite, Yahoo!, and WebCrawler. As the results come back, WebSleuth analyzes each document and URL, filters out duplicates, and assigns relevant items to the Table of Contents view and irrelevant items to the Discards view. The relevant documents are cross-indexed for keywords and phrases, and when you click on a document in the Contents view, WebSleuth instantly displays an abstract of it in the Preview window. If you come across a URL you'd like to investigate, just double-click its Table of Contents entry to load it into your browser. WebSleuth can multitask, so you can scan and analyze your results as they come in.

WebSleuth's display comprises three windows: Control, Information, and Preview. The Control window, with its menu bar and tool bar, is where you specify your search criteria and configuration preferences. After you've initiated your search, you view the results in the Information window, choosing from five available view tabs: Contents, Words, Phrases, Links, and Discards.

The Contents view lists promising documents and their URLs, generally in order of relevance. If you click on the "+" next to a document's header, you can see your search terms in context. If you switch to the Preview window, you can read the abstract that WebSleuth has generated using linguistic analysis. The Words and Phrases views display cross-indexed lists of alphabetized key terms and concepts. If you select a word or phrase, WebSleuth displays a list of all URLs that contain it, and the Link view shows the relevant documents in a tree structure. You can expand that to reveal each document's URL links. At any point during your search, you can have WebSleuth analyze those links and add them to your research.

I admit it: I was surprised just how well the linguistic analysis worked. For some time I have been looking for ways to boost my HDL (good) cholesterol levels. When I searched for "cholesterol and HDL," the search engines found about 400 hits, and WebSleuth handily whittled that list down to about 70 excellent documents. One article mentioned that coffee may reduce HDL levels. (Bad.) Because WebSleuth automatically indexes keywords for search results, clicking on the Words tab in the Information window opened an alphabetical index. I clicked on "coffee," and the Preview window displayed a list of all the URLs that mentioned coffee. A double click on any document's URL loaded that document into my Web browser for my review. (I guess I'll be switching to herbal teas.)

Installation of WebSleuth was a snap. Version 1.1 auto-installed from the CD-ROM. It requires an installation of Microsoft Internet Explorer, which is included on the CD-ROM, to work best. Once loaded onto your system, however, you can use the browser of your choice. (If you use Netscape, though, you have to rely on your browser to print search results.) I visited the Prompt Software Web site and discovered that the 1.2 version had been released. When I downloaded it and ran the installer, it updated my previous installation, keeping my registration number--preferences and all.

My only gripe is that the 24-page booklet meant to serve as the user manual included no explanation of Boolean logic or search syntax. But that's a minor quibble. For anyone who wants to search the Web, WebSleuth is an indispensable tool. It saves time and reduces frustration. And at $39.99 it pays for itself the first time you use it.

The Product
WebSleuth, from Prompt Software Inc., Novato, CA (888-4-SLEUTH; $39.99)

Bob Bruce, founder of Media Lab Inc., an electronic-publishing firm in Louisville, CO

60 MHz Pentium or higher with Windows 95 or Windows NT 4.0, 40 MB available hard-disk space, 16 MB RAM for Windows 95 (32 MB for Windows NT), Internet access at 28.8 Kb or faster