Does anyone remember Jeeves?* He was the jolly butler character that
used to greet searchers at the Ask Jeeves search engine. The trick was
that you could type an actual question into the search box, and Jeeves
would present his answers on a silver (albeit virtual) platter. Google
came along, became a verb, and stole much of Jeeves's market share, and
eventually his website got rid of him altogether, excising their name
to the less personal Ask.com. Jeeves's legacy is what librarian- and
programmer-types call the natural language query:
a query in the form of a typical, conversational question, such as
"what are the consequences of
globalization?" By the magic of the internet, search engines like Ask**,
Google, and Yahoo! are be able to understand such a question and offer
back a list of results.
If you've become accustomed to doing natural language searches (even if you didn't know you were doing such a thing), you might have experienced some disappointment when trying to do a similar kind of search in the kinds of search engines one can access at libraries, though in libraries, they are called databases rather than search engines. If you were to search "What are the consequences of globalization" in Google, you will get over 2,300,000 results.*** If you were to search the same question the same way in Academic Search Premier, one of the Roosevelt University Library's most-used databases, you will get the following message, in red letters: "No results were found."
The reason that query won't work in Academic Search Premier is because that database tries to find an exact match for what you enter in the keyword box. What natural language search engines do are strip away irrelevant terms such as what, are, and the, and search only the most relevant terms, in this case, "consequences" and "globalization". This phenomenon leads us to the true lesson of this post, one that will lead to successful searches in library databases, and even search engines like Google:
KEEP IT SIMPLE, SEARCHER!
Databases like Academic Search Premier work best when searching simple keywords and phrases. For example, if we drop "what are the" from our original search and search "consequences of globalization" in that database, you will get 95 results. The reason we get results this time is because the database searched the information attached to the articles it indexes--in this case, the title of the articles, the subjects attached to them, and the abstracts or summaries given for the articles--and was able to match the exact phrase "consequences of globalization" in 95 different articles. If we place "globalization" in one keyword box and "consequences" in another keyword box, we get 599 articles, which is a lot, but still nowhere near the 2.5 million at Google.
As you can see, the simpler your search, the higher your number of results, though you lose a bit of relevance when your results reach the hundreds. To make your searching even more effective, pay attention to the kinds of results you get from your search. Watch the subject headings that are attached to the most relevant results, and replace broad terms such as "consequences" with more specific keywords. For example, when looking at the list of 599 results from the last search, you would notice that Academic Search Premier lists "social aspects" as a subject heading. If you replace "consequences" with "Social aspects," you get 388 articles. If you replace "Social aspects" with "Economic aspects", you get 408 articles. In this database, you can click a link at the top of your list of results that says "Academic Journals," and that will narrow the list down to only articles from scholarly, peer reviewed journals--the kind of high-quality information that often cannot be found for free on Google.
Natural language continues to be the "killer app" of search engine and database searching. Lexis Nexis Academic even went so far as to make natural language searching its default search option when it changed to its new interface, but decided to switch back to a simple term search due to problems with natural language searching. Hype is brewing among programmer and librarian-types about a natural language search engine called Powerset, which no one really knows much about at the moment; in fact, it can't even be used yet.
In the meantime, remember the lesson of this post: Keep it Simple, Searcher. Until natural language becomes all it's hyped-up to be, using combinations of simple keywords will make sure your search is a successful one, no matter if you're searching the library's databases or Google.
* If you're curious what really happened to Jeeves, find out here. If you're curious where Jeeves came from in the first place, check this out.
** Those aforementioned librarian-types will be quick to tell you that Ask Jeeves was never truly a natural language search engine; enquiring librarian-types can learn more here.
*** The discussion of whether any of those results could be considered useful for research will have to wait for another day.