Yahoo, InfoSeek and Excite provide indices that organize information by subject-matter. A subject-matter index is useful when the researcher is starting his or her research and is not yet looking for particular documents or documents on particular issues. These indices allow the researcher to browse through the categories to get an idea of what is available, and then move to a search engine for more in-depth research. Because listings in the subject-matter indices are manually compiled, the pages may be more targeted, but fewer results are generated. The Spider's Apprentice provides additional information on how to navigate a subject-matter index.

Tree Research

"Tree" research can be a highly effective means of targeting relevant information. With this approach, the researcher begins with a web page or site containing relevant information and open the links on that page. In other words, someone may have already done the necessary research, and provided links to similar pages and more information. It is easy, however, to get "lost" when doing this kind of web research. One way to keep track of past sites visited is to use the "Bookmark" or "Favorite Places" function on the browser. For example, a researcher might save the URL (the web address of the page) of an interesting page as a bookmark or favorite. This ensures that the page can be found again at a later point.

Keyword Searches

Altavista, Google, and MetaCrawler are all types of engines that allow keyword searches. Keyword searches can be either simple or Boolean. Simple searches are searches for word occurrences. Boolean searches allow the researcher to specify relationships between words.

Simple searches look for occurrences of specified terms. A simple keyword search for "women domestic violence albania," for example, may pull up a website for an asylum program in Chicago that helped two different asylees, a man from Albania and a woman from Chile fleeing domestic violence. There are, however, a number of strategies that can be used to narrow search results.

It can be useful to use a number of different terms that might be included in the web pages that are sought. The researcher might try to imagine what words would be located in these web pages—for example, a page describing grants that are awarded to organizations doing work on domestic violence issues in Albania—and then enter as many of those terms as possible ("domestic violence albania organization grant award women international").

Enclosing phrases in quotation marks can also narrow results. For most search engines, quotation marks tell the engine that the terms contained within the marks are a phrase. So, instead of searching for occurrences of the words "domestic" and "violence" on a page—which may generate an article about domestic unrest that resulted in violence—a search for the phrase "domestic violence" in quotation marks will only pull up pages in which those terms occur next to each other. Not all search engines understand quotation marks as indicating a phrase. Such engines often have a separate "phrase" field—by typing two words into that field, the search engine will look for those words together.

(As the Spider's Apprentice explains, more and more search engines are automatically applying "adjacency" principles. That is, many engines now assuming that if two words are entered next to each other, the person is looking for pages in which the words are near each other.)

It may also be useful to try "wildcard" searches. The particular kinds of wildcards allowed vary with the engine used. One common example of a wildcard is the "*" sign (sometimes engines use "$"). For example, a search for "fund*" will return "fund," "funder," "funding" and all other variations. Some engines allow the use of wildcards in the middle of words (i.e., "f*nd" will generate both "find" and "fund"), but others do not.

Capitalization can also narrow or widen searches. Generally, lowercase search terms will generate both lower- and uppercase results. Uppercase search terms, however, return uppercase results. Most search engines will also pull both singular and plural occurrences of the keywords, but a search for a plural will only find plurals.

Searches can be narrowed by entering the most important search terms multiple times. In AltaVista, for example, the search "violence violence violence domestic kazakhstan" yields pages that are focused on domestic violence but mention Kazakhstan, while the search "kazakhstan kazakhstan kazakhstan domestic violence" yields pages that are focused on Kazakhstan but mention domestic violence.

Searches can be broadened by using synonyms, similar words, and multiple spellings of words. For example, the search might include both "colour" and "color," or both "foundation" and "funder." (It is possible to enter multiple spellings and synonyms all at once; it is not necessary to do separate searches for each.) Some engines have a "thesaurus" that can suggest common synonyms for search terms.

Search engines that support Boolean operators can yield more targeted searches. Boolean searches specify relationships between words. For example, it is possible to search for pages containing two words (using "AND"), either of two words (using "OR"), or one word and not another (using "AND NOT"). Most importantly, Boolean search engines also allow what are called "proximal locators" such as "NEAR" or "ADJ"—commands that allow the researcher to specify that particular words should appear close together on a page. Thus, while a simple keyword search for "women domestic violence albania" may generate all pages that contain these four words, the following Boolean search (on an engine that allows such searches) may be more targeted: women AND "domestic violence" NEAR albania. This will generate pages in which the phrase "domestic violence" appears near the word "albania" and that also contain the word "women."

(Some search engines use "+" for "AND" and "-" for "AND NOT." Often, there can be no space between the sign and the word, as in "+domestic" or "-domestic.")

These strategies, and many more, are discussed in greater depth in Terry A. Gray's How to Search the Web: A Guide to Search Tools, and David P. Habib and Robert L. Balliot's How to Search the World Wide Web: A Tutorial for Beginners and Non-Experts (1999). A more in-depth discussion of refining simple and Boolean searches is available through the Spider's Apprentice. Berkeley's Search Strategies can help researchers analyze the best approach based on the information sought (i.e., personal names, information with non-unique terms).