Let’s face it. Search is an integral part of what we do everyday and for the most part it’s really quite good. So why bother looking for a better solution? After all, learning something new is bound to be painful.

There isn’t a lot of good research data out there, but having combed through what there is it appears as if, on average, people formulate three distinct queries each time they search for something. Remember though, this is an average, so for the most part (probably 80% of the time) we find what we’re looking for in the first query. Something like “theater tickets in New York” is probably going to get you good results right away. But what this means is that the rest of the time (the remaining 20% of our searches) we actually use well over three different queries before we get what we want.

Even when we do get a query “right”, we often have to look through a half dozen pages or more to see which is the best result. And if we’re still trying to “guess” the right search terms, we hunt and peck, trying literally dozens of different pages (occasionally even looking at the second page of search engine results) before we give up hope and move on to the next set of keywords.

There are fundamentally three big problems with current search, which largely stem from the fact that conventional search engines match words and phrases not meanings.

First of all, there is the issue of ambiguity. When you type in “Java” do you mean “the programming language”, “a region in Indonesia” or “a cup of coffee”?

Second of all, you need to guess which search terms to use – you’re looking for a new “car” but it’s also referred to as an “automobile” or a “vehicle” etc. This is in fact the single biggest problem because there are sometimes dozens of different ways of saying the same thing.

And lastly, it’s the sheer volume of results that are daunting. A search in Google for “java” yields 412 million results. They can’t all be good. A lot of this is just noise but you often have to sift through mismatched or unacceptable results to get what you want.

So these then are the first things that semantic search should seek to improve. There are different ways of removing ambiguity – one being by looking at the context around a word or phrase and another to give the user a means of easily being unambiguous.

Understanding that there are different ways in which people can say the same thing is critical for semantic search. You shouldn’t have to guess the way that the search engine wants you to express yourself, you should have a search engine that understands you the way you naturally think and should do the “heavy lifting” of finding all of the results that match.

Unfortunately, the last problem, that of “noise” is pretty hard to solve. There are some trillion odd pages that currently make up the internet and sorting the wheat from the chaff isn’t easily accomplished by machines. However, with semantics it is possible to filter out pages that clearly don’t match the specific meaning of certain search terms.

So, the short answer as to why you should consider alternatives to conventional search is that you will be able to eliminate or reduce ambiguity, search using your own vernacular and not waste as much time sifting through countless pages which aren’t relevant.

In reality of course, it’s not quite as that simple. In order to semantically understand what a web page is about is not a trivial matter. But that is the subject for another day. For now, it’s enough to realize that the world could be better. After all, that’s how progress happens.