Search and Semantics: What It MEANS For You

Posted by Steve Kleber on Mar 26, 2012

Welcome to the Kleber & Associates blog! Here’s the latest on Search and Semantics: What It MEANS For You

Semantics. It’s the study of meaning. And it’s the future of search. As humans we have an intuitive sense of what semantics are. Every day we communicate by stringing words together into sentences to create and send a message. We know that individually, the words have little meaning, but when combined with other words they can hold tremendous meaning. And when we change just one word in the sentence or the order of the words, the meaning can change drastically.

Look at the following two sentences:
1. It’s time to hammer.
2. It’s hammer time.

When you read the first sentence you probably thought of a hammer, nails and a piece of wood. When you read the second sentence, however, you probably envisioned MC Hammer dancing in his signature pants. Semantics. Whether or not you knew it before, you know what semantics are.

Computers, however, do not know what semantics are. They don’t understand semantics because they don’t understand meaning. To a computer every input and output is just a string of ones and zeros.

Here are the same two sentences from above, but this time how a computer interprets them:
1. 01001001 01110100 10010010 01110011 00100000 01110100 01101001 01101101 01100101 00100000 01110100 01101111 00100000 01101000 01100001 01101101 01101101 01100101 01110010 00101110
2. 01001001 01110100 00100111 01110011 00100000 01101000 01100001 01101101 01101101 01100101 01110010 00100000 01110100 01101001 01101101 01100101 00101110

This is essentially what search engines have always boiled down to: finding the best match between the input binary code and all of the documents on the internet. As technology has progressed, programmers have worked to make search engines better “understand” what we mean by the words we use in our searches. But there’s still a lot of work to be done. While the example I used above is quite silly, it demonstrates my point quite nicely. If you type in the two sentences into a search engine you will get very similar results, because the search engine doesn’t understand how changing the order of the words changed the meaning and connotations of the sentence. Similarly, typing “hammer” into a search engine will deliver a variety of results from business websites to articles about MC Hammer.

One of the best examples of semantics and computers is demonstrated in Watson, IBM’s genius computer that defeated Ken Jennings on Jeopardy! One of the biggest hurdles that IBM had to overcome in creating Watson was getting the computer to understand the question being asked because Watson, as a computer, cannot interpret semantics. The success of Watson on the game show proves that computers are closer than ever to being able to understand and interpret semantics.

IBM isn’t the only one working on solving the semantic problem. Google is reportedly working on semantic search technology. What might these search results look like you ask? Lucky for us there are already a number of search engines out there that employ semantic search. The screen shot below is from A search for “hammer” delivered a number of possible meanings that I could be searching for. I highlighted two to demonstrate how powerful the technology is: a hammer used for construction and MC Hammer. (click image to view larger)

Subsequently, I can click on whichever meaning I am searching for and receive a list of search results about that one meaning of hammer. Rather than a Google search where you have to sort through the results by hand to find what you are looking for, semantic search is able to organize the results according to the various meanings of the key term.

If you have an iPhone 4S, you’ve also experienced semantic search. Siri relies heavily on Wolfram Alpha (along with Google), a semantic search engine, to interpret questions and deliver accurate results. Considering this, semantic search seems inevitable for Google if it wishes to stay competitive. Combining this with social search will create a very powerful tool for all of us that should deliver extremely relevant, meaningful information.

This entry was posted by Steve Kleber on Monday, March 26th, 2012 at 7:36 am and is filed under Research, Social Media. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.