EngineeringINFO Tutorial

One of the basic problems with searching for information in a very large information space such as the World Wide Web is that there are a very large number of pages that match most terms you might try. So when the total number of pages your search engine knows about is in the billions, it is not uncommon for a search engine to return over 100,000 matches for a search.

There's one big problem with this: Who wants to go through 100,000 web pages? And what makes you think that the first 20 pages are the ones most likely to be what you want out of the 100,000 pages? It is true that some search engines order their results so that the pages it calculates to be the most relevant are first - at least, after the ads - but there's a strong possibility that what the search engine thinks is relevant is not what you wanted at all!

So how do we limit the number of matches to something more reasonable? Boolean logic is a very powerful tool in this situation. Basic Boolean logic has 3 operators that help us: AND, OR & NOT.

Suppose you want to find a hotel room for a planned trip. You could search the web for "hotel" and find hotels on every continent - not exactly what you need! But how can you limit your search to hotels in Atlanta, Georgia? By using Boolean logic to say that we only want pages that mention "hotel" AND "Atlanta" we get a much smaller set of pages.

So limiting the search significantly reduces the number of pages we need to look through to find what we want. But suppose that this set is still too large? How could we limit it further? Many times your best clue comes from looking at the pages that are not what you want. For example, if your search turns up hotels in cities named Atlanta, but not Atlanta, Georgia, you would want to eliminate all the hotels that are not in Georgia. How could you use the Boolean AND to eliminate those from the set returned by the search engine?

One possibility would be to search for "hotel" AND "Atlanta" AND "Georgia" which would give us the following:

Notice that in common English speech and writing, we use "and" to include larger areas and more things. For example, when a student says, "I like movies with Tom Selleck and Halle Berry." we correctly interpret that to mean a larger number of movies than either of those stars has made individually. But in Boolean search logic, using "AND" actually reduces the matches found. So in this sense, the student would only be including movies that have both Tom Selleck and Halle Berry - a pretty small number of movies. This is because using "AND" in searching for items means that all matching items must have all the terms joined by "AND," instead of having any one of the terms. Since this is different from our habitual English usage, this is often a source of errors.

It is still likely the red area that represents the result of our search so far still contains many items that are not what we want, even though they do contain all three terms.