Different engines search in different ways. Some engines search only sites that have been registered with the engine or manually compiled (and different sites have fewer or greater numbers of entries). Others, like Google, search all pages that are available on the net. Some engines search using other search engines. MetaCrawler, for example, sends your query to a number of other search engines, like AltaVista and Google, and generates a list of the best responses. Different search engines also vary in the depth to which they will search a page.

Some search engines do not automatically look for pages that contain all of the terms entered. Rather, such search engines often seek those pages that "best" match the query. The "best" matches are those that contain all of the search terms close together and early in the document. (These engines also rank the results in terms of these same criteria, from best to worst.) As a result, however, not all of the search terms may be included in every result. In order to further narrow the results generated, it may be useful to try the "advanced search" option. This option allows the researcher to specify terms that "must" be included in the page.

Finally, different engines rank the results in different ways. Some rank by the frequency with which the search words occur, some by overall popularity, and some by the nearness of the search terms to the beginning of the page. The most relevant pages may not necessarily occur at the top of the results list generated by the engine.

Search Engine Descriptions and Evaluations

Descriptions of the advantages and disadvantages of different search engines are available from the Spider's Apprentice, Berkeley's Digital Library, the University of Adelaide Library's search engine guide, and OneNorthwest's Using Search Engines Effectively. The Spider's Apprentice also recommends Search Engine Watch for search engine listings and useful web searching tips. A few engines are described briefly below.

The WWW Virtual Library has a subject-matter index specifically targeted to resources in Eastern Europe.

Google is one of the most widely-used search engines because it is easy to use, fast, and generates relatively accurate results.

WebCrawler, particularly its "Advanced Search" page, allows complex Boolean searches (including AND, OR, AND NOT, and NEAR). WebCrawler also allows an "ADJ" command to specify that words should be adjacent to one another. This engine apparently allows a "NEAR/n" search. Using "NEAR/n"—as in, for example, "NEAR/5" or "NEAR/10"—generates results in which the two words are within, for example, five or ten words of each other.

AltaVista searches the entire web, and allows AND, OR, NOT and NEAR searches.

Excite uses "concept-based" searching. As the Spider's Apprentice explains, concept-based search engines try to determine what the researcher is searching for and to generate those results. If, based on the keywords entered, Excite determines that the researcher is searching for those words in a particular context, it will exclude results that use those words in another context.

Hotbot is a search engine that allows searches of particular domains or web pages—searches of, for example, all web pages of organizations in North America. At the bottom of Hotbot's "advanced search" page, it is possible to specify the "location/domain" to be searched—for example, only web pages ending in ".org" (North American organizations) or European web pages. Typing in a particular domain (i.e., mnadvocates.org) will allow a search of the web page of MN Advocates for Human Rights. The "http://www" does not need to be included when searching one particular page. It is also possible to limit the search by date and designate the depth of the search (i.e., search only the top page, or only the first two levels of pages).

Two search engines that focus on women's issues are FeMiNa and WWWomen. Each provides a subject-matter index and keyword search capability. These engines appear to only search registered entries, and the results therefore may be less targeted than results generated through non-registration engines.