The Art of Effective Web Searching
Six advanced search techniques that will help you find the best results faster.
The ubiquitous Web search form (the“search command line”) has had aprofound and transformative effecton information retrieval. That simple littlebox at the top of every search engine hasopened up a boundless new world formillions of people. In doing so, all of themajor search engines have made compromisesto cast their nets wide and broadentheir overall appeal. While this has donewonders to build a general-purpose tool forthe majority, there are specific things thatwe can do to unlock the secrets buriedbeneath the search command line.
The major search engines tend to returnparticularly accurate and relevant results fortwo very different types of queries. On onehand, a search for terms that are commonbut rather specific will usually return anaccurate list of popular Web sites thatcontain canonical information about theterm(s). At the other extreme, a search for arelatively rare and statistically improbableset of search terms also tends to return accurateresults. For example, a search for “ISBN0679723420” points you straight toVladimir Nabokov’s classic Pale Fire,without much effort.
Therefore, the art of effective searchingusually falls somewhere in the middle ofthose two kinds of queries. This is especiallyapplicable in the academic and technicalcommunities, where users frequently seekout esoteric and difficult-to-articulate data.While the search engines are getting better athelping the user with these more demandingqueries, here are six techniques the poweruser can adopt to find the best results fast:
Search from the outside in. Sometimessearch queries simply return too much information,especially when the search engined'esn’t know enough to disambiguatebetween different types of results. The firstapproach most people try is to add moreterms to the query until the right results arefound. But this can just as easily eliminategood results as reveal them. An alternative isto start with broad search terms and use the“minus” operator in the search query toselectively eliminate sets of results. This letsthe power user quickly scan through the listof results and subtract entire categories ofpages with quick modifications to the query,ending up with only the desired matches.
Comparison search. The Web is so large,no single search engine can crawl more thana fraction of it. Also, each search engine’srelevancy algorithms (i.e., the logic thatdetermines the order of the search results)differ substantially. The solution is to keepbookmarks handy for the big engines such asGoogle (www.google.com), Yahoo! (www.yahoo.com), and MSN (www.msn.com),and be prepared to check all of them onparticularly challenging searches.
Server-side bookmarks. Everyone usesbookmarks, but one power search techniqueis to use a server-side bookmark managerrather than the one stored on your computer.When the user finds great pages, the bookmarkscan be saved on a dedicated bookmarkserver such as del.icio.us (del.icio.us) orLookSmart’s Furl (www.furl.net), or one ofthe advanced search engines such as A9(a9.com) or Yahoo!, via their toolbars. Manyof these new services also support the capabilityto tag individual bookmarks withkeywords or add extended diary entries thatannotate your personal Web as you search.
Tabbed browsing. Web browsers such asFirefox (www.mozilla.org/firefox) andSafari (www.apple.com/safari) supporttabbed browsing, which hides new pages inthe background of the same window untilyou need them. This allows the power user toperform a search and preload all of the interestingresults in tabs while they work downthe list. So, all of the selected pages will beloaded and ready to be read by the time theuser is done scanning the list of search results.
Search histories. Another technique isto employ a search engine to track the user’ssearch history. A permanent, searchablehistory is invaluable when performingserious research on the Web. In fact, searchhistories are often better than bookmarks insuch cases, because a bookmark only helpsthe user remember one site at a time,while asearch history helps the user remember howand where she found the page, and helps herretrace those steps in the future.
Vertical search. No one general searchengine will ever be able to fully capture therichness and complexity of the Web. Medicaldatabases, travel information, and librarycatalogs are a few examples of vertical searchengines that contain highly specialized datain narrow domains. Search syndicationprotocols such as Amazon.com’s Open-Search (opensearch.a9.com) are helpingspecialized vertical search engines (e.g.,TheSeattle Public Library’s search engine foundat www.spl.org) surface their results to othersites to help maximize their accessibility.
The search industry is evolving rapidly,and more tools for information retrieval andinformation sharing are being developeddaily. Search engines are also learning how topersonalize results based on the user’s individualhabits. They are incorporating moreand more of the deep vertical data into theirprimary search interface, and are learninghow to simultaneously serve the noviceuser, while adding tools and capabilities toempower the most sophisticated users. Inthe meantime, power users can stay on top oftheir searches with advanced search techniquessuch as those above.
DeWitt Clinton is the A9.com softwaredevelopment engineer responsible for leadingthe search company’s OpenSearch feature, atechnology that allows third parties to maketheir content searchable and the results viewablethrough their own columns on theA9.com Web site. He has a bachelor’s incomputer and political science from WilliamsCollege in Massachusetts.
This article originally appeared in the 09/01/2005 issue of THE Journal.