UK United Kingdom

Google’s Knowledge Graph – has search just changed forever?

Late last week, Google representatives unveiled a significant enhancement to the company’s ubiquitous search engine. They’re calling it the “Knowledge Graph” and claiming it will support “more intelligent…

Semantic search is about recognising the meaning of words, not just the words themselves. Chris P Jobling

Late last week, Google representatives unveiled a significant enhancement to the company’s ubiquitous search engine. They’re calling it the “Knowledge Graph” and claiming it will support “more intelligent searching for real-world things on the internet”.

And while it might be a while before Australian users have access to the Knowledge Graph, the US roll-out is expected to begin in the coming days.

So what is it? How does it work? And will it change the way we find information online?

According to the company, the Knowledge Graph encompasses three new features:

“Find the right thing”

If your search term is ambiguous and has multiple meanings, the new interface will help to narrow your search. For instance, if you search for “Madonna” a panel pops up allowing you to choose between the religious icon and the pop star.

Click for larger view. Google

“Get the best summary”

Search results will now include additional key facts about the topic being searched. So if you search for “Frank Lloyd Wright”, Google’s Knowledge Graph knows that your search text refers to a famous person, and so that person’s birth date, death date, education, spouse’s and children’s names are all displayed.

Click for larger image. Google

“Go deeper and broader”

Search results will now also include key relationships to other resources on the web. In the case of Frank Lloyd Wright, the Google Knowledge Graph knows he’s a famous architect, so the results include links to buildings he has designed and other significant architects from the same era.

Click for larger image. Google

The aim is to provide a more intelligent search engine – one that isn’t based on simply matching strings (a sequence of characters, such as a word) to single web pages. Instead, the Knowledge Graph will “understand” what you’re searching for and provide more relevant and precise information.

Moreover, the Knowledge Graph can identify and retrieve connections to related “things”, such as people, objects, places and events. For example, in the case of Frank Lloyd Wright, the search returns links to buildings he’s designed and other famous architects from his era.

Such relationship or “knowledge” graphs provide an intuitive interface for understanding a topic, its context in the wider web (and world) and for triggering new lines of enquiry.

So how does this technology work?

About a year ago, Google acquired Freebase, an “open shared database of the world’s knowledge”. It’s a repository of structured knowledge describing over 20 million entities, each with a unique identifier, a type (people, place, book, film, building) and a set of properties (e.g. date of birth for a person, latitude and longitude for a place).

Each entity is represented by a topic node in the massive graph that underpins the database. Properties can be used to specify relationships between entities and topics. For instance:

Dr Strangelove {film} has_director {property} Stanley Kubrik {person}.

Using the property “has_director”, the above example links the entity Dr Strangelove (of type “film”) with the entity Stanley Kubrick (of type “person”). The Freebase knowledge graph is built on millions of these connections, having been contributed through crowdsourcing, much like Wikipedia. Consequently the database will continue to grow and improve over time.

Freebase’s approach builds on “semantic web” technologies and the more recent Linked Open Data movement.

The semantic web is a World Wide Web Consortium (W3C) initiative that aims to provide meaning to objects and pages on the web. This is achieved through standardised, machine-processable metadata descriptions plus persistent unique identifiers or URIs (uniform resource identifiers).

Quite simply, the semantic web is about creating a “web of data” from an otherwise-unstructured jumble of web pages from all around the world.

The Semantic Web has been driven to a large degree by Tim Berners-Lee, the inventor of the World Wide Web. While the concept of the Semantic Web has been around for more than ten years, the vision described in a 2001 Scientific American article by Berners-Lee has largely been unrealised.

In that vision, humans, computers and software agents on the web automatically and seamlessly communicate and interact together intelligently. In the article, Berners-Lee gives the example of a semantic-web-enabled microwave oven consulting the frozen food manufacturer’s web site to retrieve optimal cooking parameters.

Many see Google’s “Knowledge Graph” as the coming-of-age of the semantic web. Finally, semantic web technologies are validated by their underpinning of the world’s biggest search engine, used by millions daily, to provide intelligent, contextualised search results.

So where is this heading in the future?

As mentioned, the Knowledge Graph on Google Search will be initially rolled out only in the US but other countries, including Australia, will likely follow (depending on the popularity of the service).

According to Google, a tailored version of the Knowledge Graph for smartphones and tablets will also be available soon.

Only time will tell whether users love or hate the new semantic search features. But even if users don’t like it now, it’s highly likely that semantic searching will be a feature of search engines in the future.

Over time, the data underpinning the Knowledge Graph will grow, enabling even smarter, richer search results. In addition, Google has the huge advantage of being able to combine its existing data on who searches for what, with Freebase’s massive corpus of linked open data, to identify the most sought-after facts or relationships that should be recorded for each entity.

This combination will lead to the killer Google app – a search engine that delivers both personalised and contextualised semantic search results.

Sign in to Favourite

Join the conversation

9 Comments sorted by

  1. David Glover


    Just correcting an error: Tim Berners-Lee invented the World Wide Web, not the Internet.

    1. Matt de Neef

      Editor at The Conversation

      In reply to David Glover

      All sorted David. Thanks for the tip-off.

  2. Dale Bloom


    Another article on software from an overseas company, which inevitably encourages people to import software.

    How about an article on a "search solution that searches content across Internet, intranet, database and shared network drives, as well as electronic document management and portal systems."

    And was invented here in Australia.

    1. Jane Hunter

      Director of eResearch, School of ITEE, Uni of Qld

      In reply to Dale Bloom

      You're missing the point - Google is the worlds biggest search engine by a large margin:

      When Funnelback has a billion users per month and processes over 1 billion search queries per day (including many from Australia), then I'll happily write an article on it.

    2. Dale Bloom


      In reply to Jane Hunter

      Google may be the biggest search engine in some areas, but not necessarily the most efficient. My point is that so many articles are being written on foreign software, which makes it more difficult for local software developers to operate.

      Meanwhile there are more people wanting Australia to have a diversified economy. I think they could rule out IT as being part of that diversified economy.

    3. Patrick Sunter

      PhD Candidate at University of Melbourne

      In reply to Jane Hunter

      Well I'd agree Dale has a point also - while software is inevitable a global industry, as an Australian website it would be good if writers on the Conversation did put some focus on local innovations and developments. Plus if we only ever wrote articles about the 'big boys', that would be a self-reinforcing phemonenon, no? For example, as the article states, Tim Berners Lee's "semantic web" concept has been developed from a small base for more than a decade, including via academia, and it's only now that large corporations are taking it up and implementing it.

      And not in any way wishing to recommend a commercial product, but another Australian innovation in this space was in text-based database search developed by researchers at RMIT and elsewhere, which later became the (internationally-used) commercial product TeraText.

  3. Patrick Sunter

    PhD Candidate at University of Melbourne

    This is certainly interesting news, and good that Google's using the latest technology to make searches more relevant.

    Apropos to this, a powerful point was made by Eli Parisier recently at a TED talk that over-personalisation of online content by Google etc that learns our 'preferences' (including habits & biases) has both advantages but also social problems - see .

    Parisier and others have just launched a new social media project, "Upworthy" (, to try to help & encourage us to share more 'meaningful' things online - including those that challenge our world-view sometimes.