Google Oils Engine With Semantic Search

Google on Tuesday rolled out two enhancements to its search enginethat could pave the way for a shift toward more integrated real-timesearch results. The new technology introduces new algorithms that draw from semanticrelevance of search terms.

The first enhancement expands the list of useful related searches. Thesecond enhancement lengthens search result descriptions, called”snippets.” Both new offerings are designed to help guide users moreeffectively to the information they need.

“These changes are highly significant for people who don’t find whatthey want in a search. We’ve been observing searching behavior andfound that many users get poor search results because of the searchwords they use,” Ori Allon, technical lead for the search quality teamat Google, told TechNewsWorld.

Mind Reading

Up until now, Google could not help userswhose keyword selections were too broad, yielding them poor results, saidAllon. Their search words were not optimized.

The new technology tries to interpret the search terms. Making thesnippets longer will help people find more quickly the informationthey are looking for, Allon said.

“If a query has three or four words, it is hard to describe theresults. We saw this as a better way to describe the results,” heexplained.

How It Works

The new technology is based on algorithms Allon developed beforecoming to Google in 2006. Instead of using search histories thatGoogle tracks, this new approach is strictly algorithm based and isautomated and scalable. The new features, which are based onproprietary software, support 37 languages.

The new method allows Google to quickly search in real time afterconsidering the semantic significance of the search terms users enter.The algorithms apply associations and concepts related to the searchterms.

This first new application within Google Search offers more usefulrelated searches in the terms placed at the bottom and top of thesearch results page, according to Allon. It can focus more accuratelyon queries and provide suggestions that are more relevant.

The technology analyzes Web page content related to the query anddetermines the relevance of people, places or concepts to the query.The process only needs fractions of seconds.

More Than Semantic

Google’s new technology is not solely based on advanced semanticmethodologies, noted Allon. That goal may be reached in the future.But semantic characteristics are just one part of this new technology.

The most popular search method is still based on keyword searchprinciples. Its results are highly dependent on how pinpointed usersmake their keywords. Often, useful information is buried among thousandsof hits that never get looked at in detail.

Google plans to integrate more features related to this new searchingtechnology, but Allon declined to provide details or a time line.

Initial Reactions

Google’s market share continues to increase because the company isable to stay on the cutting edge with its technology, noted ScottTesta, professor of marketing at St. Joseph’s University.

“The idea is, by having semantic results the user experience willimprove. People will better find what they are looking for,” Testatold TechNewsWorld. “The bottom line is that Google generates so muchcash from searches. By offering this enhancement, Google will stay inthe lead.”

Relevance Counts

The enhancements are very significant from a search engineperspective, said Heather Lutze, the founder of search enginemarketing firm Lutze Consulting. Google is providing clean resultsbased on surrounding words. The technology is looking for commonalityamong search terms, she said.

“The search results will be higher value because they will filter outthe junk. The result is trying to get people to care about their userexperience,” Lutze told TechNewsWorld.

The new Google search enhancements signal a step toward the future ofsearch, and it will dramatically challenge Microsoft and Yahoo, addedKelly Cutler, CEO of Chicago-based Marcel Media.

“Google’s semantic search will lead to more relevant results. It makessearch smarter, understanding associations and concepts related tosearch terms so that a user can search with words typically used inconversations. It will eliminate low quality page links and Web sitesfrom search,” Cutler toldTechNewsWorld.

1 Comment

  • We’ve been doing similar things at Duck Duck Go (http://duckduckgo.com/) to these new Google changes. In particular, we put zero-click info, e.g. topic summaries, on top of links. We also also put an explore box with related topics above links. For example check out http://duckduckgo.com/?q=futurama

    Our snippets and related topics are not algorithmically driven, however. Instead they are based on human edited sources, e.g. Wikipedia and Crunchbase (and many others). Consequently, they are more relevant and make more sense than Google’s info. We also have more semantic properties, such as ambiguous keyword detection, e.g. http://duckduckgo.com/?q=apple, which of course Google does not.

    See http://duckduckgo.com/about.html for some more examples. Of course, we’d love your feedback on what we’re doing.

    Gabriel Weinberg, Founder & CEO

Leave a Comment

Please sign in to post or reply to a comment. New users create a free account.

More by Jack M. Germain
More in Search Tech

LinuxInsider Channels