Semantic search is an SEO industry buzzword that not too many people fully understand. On the surface it’s pretty easy to figure out: it is search based on understanding the context and associated meanings of search queries. It’s supposed to deliver highly relevant results and take the frustration out of searches that go nowhere – searches that require retyping and rephrasing and travelling from one search engine to another in the hope of better results.

Wikipedia cites a paper by R Guha, Rob McCool and Eric Miller that delves into the intricacies of semantic search. In the paper, the authors define two main types of search: navigational search, which has a specific purpose, e.g. to find a particular article or website; and research, which has no specific end goal but seeks to amass information. Semantic search aids research, but doesn’t have a big role to play in navigational search.

Google has been trying to get into semantic search for years. That’s why you get all those suggestions when you start typing in a query. It doesn’t always get it right. In fact it can be down right annoying when it tries to give you results before you’ve finished typing your query. But it is trying and it is getting better.

Microsoft has tried to bring semantic search into Bing. A couple of years ago it acquired a semantic search engine called Powerset, which it integrated into Bing’s algorithms to deliver semantically relevant search results. It’s not a bad effort. In my opinion, it’s a better effort than Google. But it’s not perfect.

For proper semantic search results you need to go to a proper semantic search engine, like Hakia. There are quite a few self-styled semantic search engines, some of which have decided to specialise in specific niches. My favourites are the above mentioned Hakia and Exalead. The nice thing about Exalead is that it breaks up its semantic search into images, videos, labs and blogs.

What do semantic search engines do differently to other search engines?

According to Hakia, semantic search engines have to be able to:

• Determine morphological variations – tenses and plurals.
• Determine applicable synonyms.
• Cut through generalisations.
• Match concepts and knowledge.
• Answer questions.
• Highlight the most relevant paragraphs and sentences in results.
• Negate the need for the special characters that narrow down traditional searches, like inverted commas.
• Not base results on potentially misleading, artificial information, like previous searches.
• Recognise their own shortcomings and make appropriate adjustments.

Even with all their specialisations, semantic search engines are still not perfect. In the online world few things ever are. The important thing is that the people behind the search engines know this and work constantly to improve upon their performance.

The result is that semantic search is not a traditional search killer; instead it complements it rather nicely.

(Image by dullhunk, CC by 2.0, via Flickr)