www.PlanetIT.com The Network For IT Professionals

Speeding The Information Hunt
by Greg R. Notess (11/05/98; 9:00 a.m. ET)
URL: http://www.PlanetIT.com/docs/PIT19981110S0009

With so much IT information on the Web, what's the most effective way to find a quick answer to a question? Several techniques can make finding the right Web page easier. No single technique answers every question - but an intelligent combination of savvy searching and picking the right technique helps expedite the Web search process.

Using the capabilities of phrase searching, available on most popular search engines, is one of the best search techniques for finding relevant hits amidst the mass of unrelated pages retrieved on most typical Web searches.

Phrase Searching
Searching a single word with any of the Web search engines can find thousands if not millions of pages. Yet add a second word in the proper way, and search results can suddenly become much more relevant.

Telling the engine to search only for pages that contain both the words can be done, sometimes, by joining the two words with AND or by putting a plus sign directly in front of each of them. While this Boolean "ANDing" technique significantly reduces the numbers of hits, phrase searching does even better.

Fortunately, all the major search engines use the same syntax. Designate a phrase by surrounding it with quotation marks. This tells the search engine to find pages that have those exact words, in the exact order specified, and immediately adjacent to each other. For example, a search statement of "this exact phrase" should only find pages that include the phrase "this exact phrase."

Here's an example: A person searching for a list of commands used in Hewlett Packard's MPE/iX operating system might try a search on AltaVista using just MPE/iX.

This finds over 2,400 pages. An AltaVista advanced search in the Boolean query box of MPE/iX and commands finds more than 450. But a phrase search on "MPE/iX commands" retrieves only 15, including an Idaho State University Computing and Communications Administrative User Guide with an elaborate section on MPE/iX system commands.

Determining Phrases
For general Web searching, using a phrase search to narrow results is an excellent way to get to more relevant search results. However, using phrase searching well also involves coming up with a phrase that would find pages to answer the specific question. Sometimes, simply entering a question as a phrase may work, especially on AltaVista, now that it has adopted the Ask Jeeves search engine, which identifies common questions and gives preset answers.

The phrase search "why is the sky blue" on AltaVista finds an answer in the Ask Jeeves portion but also identifies several pages that also try to answer the question.

However, the straight question approach only works for a limited set of searches. Other phrases also can work quite well, especially when one of the words in the phrase is somewhat unique or the particular combination of words is unique. Try a phrase search for people's names, product names, or even an embedded programming string.

Receive an unknown error message from a particular program? Try searching the error message (or some significant portion of it) as a phrase. When daylight-saving time changes, Netscape Navigator can start producing an error message containing the phrase "bookmarks have changed on disk." Search the phrase on Northern Light or HotBot to find several pages that explain the problem and a fix for it.

The DejaNews Exception
For searching error messages and other software discussion, DejaNews and its archive of Usenet postings often can find answers for problems that may be difficult or impossible to find on general Web search engines. Since Usenet is more focused on discussions while the Web is generally more static, DejaNews' Usenet archive may also be a quicker source.

Unfortunately, DejaNews does not always process phrase searches fully. It accepts the common syntax, but its documentation notes that phrase searches are processor-intensive and will often only give partial results for a phrase search.

For this reason, on DejaNews, it may well be preferable to just enter the phrase without the quotes and hope the hits that do contain the phrase rise to the top. Some of the Web search engines process phrases in unexpected ways - ignoring stop words, searching alternate forms of the word, or dropping spaces -- but none seem to have the processing problems of DejaNews.

Going To The Source
Sometimes phrase searching is too precise or a single phrase may not accurately represent the information need. Another approach is to go directly to a likely source. This technique is almost too obvious to mention - but it works well. Rather than using Web search engines at all, just go to the Web site of the organization most likely to provide a particular kind of information. Much of the time, the URL can be guessed using the www.companyname.com formula. If it cannot be guessed quickly, then try a directory like Yahoo! or a search engine.

For example - to find specifications on accessories available from Dell, try the Dell site at www.dell.com. Wondering what's new at Oracle? Try www.oracle.com and look in their news section. An obvious strategy, but one that is often forgotten in the time spent pulling up irrelevant information on the big search engines.

This technique can be used for other searches, not just company-related queries. Government agencies, professional associations, and educational institutions can all offer excellent informational resources. For population statistics, go to www.census.gov or use www.acm.org to see a list of upcoming conferences.

New Search Engine Features
In recent months the search engines have been busy adding improvements to help make finding relevant material easier. And there have been some significant improvements. As mentioned earlier, AltaVista incorporated the Ask Jeeves technology in its simple search. If your question matches one of the commonly asked questions in the Ask Jeeves database, it can be a quick and easy resource.

This works well for questions like "What time is it in Paris?" but not as well for questions such as "What is the best way to make an Oracle database accessible on the Web?"

AltaVista also has added the RealName system, which provides links to the websites of registered companies when the company name or one of its products is searched. HotBot uses Direct Hit, which gives a select list of up to 10 hits most frequently chosen by others who search the same words. Both of these approaches can be quite effective for finding an organization's site when you cannot guess the URL.

Excite takes an even more sophisticated approach these days, giving links to a company's website, its address, current stock price, news articles, and more. For the topics like major company names that Excite can correctly identify, this is an excellent way to get basic information and current news. However, it is much less successful with more detailed questions.

These techniques are just some of the best ways to speed the information hunt on the Web. Phrase searching, going direct to the source, and taking advantage of the newer output features of the search engines can all make for more effective information retrieval.

About The Author:
Greg Notess is reference librarian at Montana State University. He is an author, lecturer, and speaker on Internet search issues.

His book, "Government Information on the Internet" (Bernan, 1998), is in its 2nd edition, and his Internet Access Providers: An International Resource Directory is the de facto guide on the topic.

Notess also writes columns for publications such as Online, Database, and the Internet Law Researcher; and he speaks at a wide variety of Internet conferences across the country.

www.cmpnet.com The Technology Network

Copyright 1998 CMP Media Inc.