The goal of any search is unified retrieval. The retrieved results should not be fettered by artificial means -- technology, access methods, or formats -- but rather, limited only by the knowledge, skills, and experience of the inquirer. When constructing a query, the technology often reduces our search capabilities, forcing us to describe our query as a single word, a word combination or a phrase. However, to achieve unified retrieval, we, at least, must view the entire object as a query and we must wholly, and accurately, describe that object. One object has one description. For example, a photograph of a work of art is actually two objects and therefore, requires two descriptions. The photograph itself is one object; the work of art is another. If you are searching for the Mona Lisa on the internet, you need to decide if you are looking for the work of art itself, or a rendering of that work of art. If you are looking for information on wolverines, are you looking for information on the animals or the University of Michigan football team? Perhaps your query needs to be structured in some way to include the relationship between the object and its description.
Dealing with vast amounts of information means constructing the best search possible to achieve the best retrieval possible. When constructing a search, especially on a system as vast and uncontrolled as the internet, you first need to select the appropriate information infrastructure. Is the object you are seeking most likely to appear on a web site, an ftp site, be contained in a UseNet newsgroup, or a specialized database? Should you restrict your search to a geographic area, a particular discipline or subject area, or a particular domain; each of these categories may have search engines which limit themselves to those categories. Secondly, you need to build a proper query. This may involve using boolean and/or positional operators, eliminating noise words, identifying alternate terms for your query, narrowing or broadening your search, and truncating and masking as appropriate. When items are retrieved, you have to consider them in light of filtering -- have you limited them as you chose, or have you placed unintentional limitations on them because of your choice of information infrastructure or search construction? Do you truly understand how the search engine works or has the technology limited the retrieval without your being aware of it? For example, if you search for grapes, do your returns include: grape, rap (the noise and the music), wrap, rape (the act and the seed), grapple, rapists, rapid (transit and water), and so forth? If so, the technology may be both left and right truncating your search to the stem rap and returning everything with that stem present. Occasionally, in some systems, you might experience customized returns. They may be customized by the technology through the use of intelligent agents, cookies, or bots, and you may be aware they are customized or you may not.
Sometimes your returns will be influenced in ways you cannot control no matter what you do to your search. Indexing by publicly accessible search engines is generally an automated process. These search engines assume the web site creator has self-described the site appropriately, through proper coding construction, meta tags and the like. If this does not occur, or if the creator deliberately tries to fool the automated indexer, there is very little your search can do about it. As well, because the indexer only presents a snapshot of a portion of the internet as it existed at one point in time, you may retrieve pages that no longer exist or have been moved since the indexer did its work.
With its overall size, its dynamic nature, its lack of controls, and its design, the internet can provide hours of wasted time when a researcher is attempting to find very specific pieces of information.
Top of Page
This page copyright, created and maintained by Linda Hansen.
Comments and suggestions to: firstname.lastname@example.org
Created: 1999/10/10 Last updated: 2010/08/19
Terms of Copyright
This document: ...search/query.htm