Web and Semantic Search

Semantic data refers to a model in which stored data is given semantics or meaning in relationship to the “real world.”

In the search engine sphere, therefore, specific keywords can be interpreted according to the reason that people search for such keywords. For example, take the keywords “Big Ben” referring to the famous clock located on the Parliament building in London. Big Ben is located in the “United Kingdom“ and in “London”, so there is a semantic relationship in terms of location.

How search engines use semantic data
Search engines try to identify “structured data” i.e., keywords that have semantic relationship to other keywords. Thus, someone searching for the key phrase “Big Ben” might also be interested in general information on the city of London, or the country of the United Kingdom.

Studies have shown that most search engine users are looking to find information for practical real time purposes. For example, most people searching for “Big Ben” might be tourists who are preparing to actually visit the site in the real world. So they are looking for practical and timely information.

The major search engines like Google, Yahoo and Bing can provide users information closer to what they are actually looking for by studying the semantic relationship between keywords (data).

Ascertaining semantic data generally requires studying how keywords are used in relation to one another during searching, or how they relate in text found on web pages. The practice of tagging web pages, blogs, photos and the like also helps in identifying semantic relationships between keywords.

Semantic search results
Although there are search engines that specialize in semantic search like Hakia, Powerset and Cognition, the popular search engines like Google have also incorporated semantic data structures into their search results.

On Google, if one submits the query “restaurants San Francisco”, a set of local business results appears at the top of the search engine result page (SERP). The local business results appear at the top because the search algorithm identifies the keywords as indicating that the user is likely looking to actually patronize a local restaurant.

A search for “Tokyo” instead will not turn up any local business results at all. Instead at the top of the first results page we have an image from Google Maps. The semantic result indicates that most people searching for “Tokyo” are interested in finding directions or else looking at a map of the city. Following the Google Map results are a set of regular search results focused mostly on tourism, and then in order: news results, video results, Google book results, and image results. From the types of links suggested, we can surmise that the structured data related to the keyword “Tokyo” is focused on users who are looking for travel information.

The semantic data model, thus, allows search engines do deliver better “real world” results to users based on the semantic relationships and data structures that apply to each keyword. Advertisers and webmasters can also improve performance by understanding these relationships.