semantic search, in-house vs webwide, findability, keywords vs semantic formats, unique name spaces, examples, future.

Semantic search will be all about findability. In the semantic future, you don’t search for anything. You say what you’re looking for and let it find you. Today, search is a hit-or-miss project of typing in keywords and looking at web sites. If you’re looking for something hard to find, you must monitor dozens of sites to see when it shows up at the price (or other conditions) you want. In the future, you’ll specify a range of conditions and then let the search engines bring you descriptions from the open web. Because the descriptions will be machine readable, software will help filter out all the incoming responses and give you what you’re looking for. The same will be true in the knowledge domain – rather than typing in keywords, we’ll ask questions and get answers back.


A great example of a transitional strategy is Wolfram Alpha, a huge effort by hundreds of people making data more understandable to machines so it can bring you answers, rather than web sites. What makes it significant is that this search engine sees a tiny fraction of the knowledgesphere, but what it can see it really knows. And people are now starting to publish their data in formats that WA can understand natively. This could have a huge impact on search eventually.

At the moment, we have semi-structured search. For real estate, you can search on Trulia.com or Realtor.com. You can specify things like zip code, number of square feet, bedrooms, bathrooms, amenities, etc. You can do a very semantic but very limited search for menu items at SeamlessWeb.com. And you can search generally using new semantic search engines listed here. All of these search engines must guess what you are looking for and what matches to some degree. They try to add structure to the unstructured information already online. Most of them aren’t very useful, but they will likely get more powerful over time.

Google Squared


Powerset (purchased by Microsoft)




Fully Semantic

We are several years away from fully semantic web-wide searches, but fully semantic searches do exist inside certain web sites. For example, see the diamond finders at Amazon.com and Bluenile.com. These tools let you search their entire inventory of loose diamonds by specifying your desired range of cut, clarity, weight, color, and shape. Imagine being able to do this will all the loose diamonds in the entire world? That’s the power of semantic search. Set your search terms and let the products, services, and answers find you.