Speed searching.(on the net)(Viewpoint essay)

In these days of internet plenty, we have a wealth of searchable databases and online content--from Google to ScienceDirect to Amazon.com and a myriad of specialized databases (both free and fee). We can search the full text within journal articles, online reference books, websites, government documents, technical reports, and many other sources.

Online searching is quick, and so many resources can be found with amazing speed compared to just a decade ago. At the reference desk, when helping someone decide which database will have the most relevant resources for a multidisciplinary topic, I test the same search at a variety of databases--commercial and on the free web. At a simple level, federated searching strives to be an answer to this type of problem. Unfortunately, for many searches, a different combination of databases is needed--sometimes including one not available within a federated search. When a federated search is too large, slow, or confusing, users turn to a professional who can quickly search several additional resources to find those special documents with just the right information. Assuming they don't simply give up on the entire process, that is.

The super searcher can quickly try a search at several databases, skim results, reword the query, and deliver relevant content. Searching quickly means faster access to information. While there are many approaches to speed searching, this column explores using short queries, long phrases, search suggestions, and search switching techniques to help users get to the information they want.

HISTORY OF SPEED

I still remember some of my online searching lessons from library school (as it was called in those days). With the range of query languages and diverse search systems, one approach I was taught was to create a query with the fewest number of characters that would still find the correct record. This is still good advice for speeding up the search process--and not just for those with slow keyboarding skills.

Back then (in the late 1980s), some query syntax emphasized easing the computer processing burden by using unique search languages. The longer the query, the longer it could take to process the search. Thus, OCLC had specialized, derived, searches, like the 3,2,2,1 title search. For anyone unfamiliar with such a search, it uses only the specified number of characters from each of the first four title words (excluding initial articles) so that a search for The Extreme Searcher's Internet Handbook would be ext,se,in,h. OCLC still supports this title-derived search, along with the 4,4 for author and title; 4,3,1 for personal name; and =4,3,1 for corporate name-derived searches. Give it a try in the expert search mode in WorldCat on FirstSearch.

Even on search systems that used full words in queries, choosing unique words minimized the number of keystrokes necessary. It can become a game to choose the most unique words that find a specific result. Back when every character in a search increased cost and computer processing time, it made sense to minimize the length of a query.