Searching 101

Computer Life column for 11/1/97 by
Richard Gordon


Looking for things on the Web has become too easy and too hard.

There are now hundreds of tools that purport to make it easy to find things on the Web, but each has its own style and criteria.

Where to begin?

Some of these search tools are like a library's catalog, a collection of links and pointers to information, all organized by subject. You can browse this kind of subject catalog by topic, or you can search for specific keywords.

Sometimes these subject catalogs are very broad and inclusive; sometimes they are very narrow, focused on only one topic. Usually, they are moderated, with someone exercising "editorial control" over what sites get listed.

The other major type of index is built by automated robots and spiders, programs that go out over the web to gather information and build a database for you to search. For the most part, these services are designed to be more inclusive. They are all keyword searchable. However, what they do not have is the ability to let you browse through a list of links by topic.

If you want to browse and poke around by topic, perhaps you should begin searching with a subject catalog like Yahoo (www.yahoo.com) or Magellan (www.mckinley.com). Both are very well organized and easy to navigate. Magellan even lets you limit your search to "green light" sites that are safe for family viewing.

If you want to do keyword searching for a specific topic, start off with one of the automated search engines. My two favorites are Alta Vista (www.altavista.digital.com) and WebCrawler (webcrawler.com). Alta Vista is the more inclusive.

Do you have lots of time to invest in your Web search in the name of thoroughness? Then begin with one of the "meta-search" sites: for example, c|net's search.com (www.search.com) or Dogpile (www.dogpile.com).

Search.com is really an index of hundreds of different search programs, topic catalogs, and indexes. You can even search the search tools to see if there is a specialized index or catalog in your area of inquiry.

Sites like Dogpile allow you to type your search one time and have it passed to multiple search tools. Dogpile is a useful application of the principle that, if one tool strikes out, thirteen others may find what you are looking for.

Narrowing your search

Once you begin searching, the problem is usually that you find too much information, and not all of it is relevant. When that happens, look at the help for the search tool and see how you can limit your search.

For example, if you search for Kit Carson, you may get thousands of hits: some with "Kit", some with "Carson", and some with both. If you are researching the historical figure, you need to narrow your search.

Many search tools, e.g., Alta Vista, let you use the plus sign, the minus sign, and quotation marks to focus your search. The plus signs indicate that a Web document must have that text on it; a minus sign indicates that you want to discard any sites with that text; the quotes indicate text that should be grouped together. For example, if you add Carson's birth and death years, put quotes around his name, and use plus signs (+"Kit Carson" +1809 +1868), you'll get much better results.

Limitations

Not everything is on the Web-yet. What you're most likely to find are article-sized chunks of information, not book-length works. For example, if you are helping a third grader find material for a report on Kit Carson, you will find plenty of information for his purposes: lots of brief, encyclopedia style articles.

However, if your tenth grader is researching the same historical figure, she may need more information. Certainly she can get an overview of Carson's career, but the questions and gaps in the short articles should help her focus her further research. Ultimately, it may be old-fashioned bibliographic references she finds on the Web that will help her the most.


Copyright © 1997, The News Journal Company

Computer Life Index

Richard Gordon helps support faculty, staff and student computing at the University of Delaware. E-mail questions, comments or suggestions to richard@inet.net, or write him at The News Journal, Box 15505, Wilmington, DE 19850. Although each note cannot be answered individually, reader comments and questions will often be incorporated in future columns.