Google Launches Knowledge Graph To Provide Answers, Not Just Links

Hinted at for months, Google formally launched its “Knowledge Graph” today. The new technology is being used to provide popular facts about people, places and things alongside Google’s traditional results. It also allows Google to move toward a new way of searching not for pages that match query terms but for “entities” or concepts that the words describe.

Knowledge Graph? “Graph” is a technical term used to describe how a set of objects are connected. Google has used a “link graph” to model how pages link to each other, in order to help determine which are popular and relevant for particular searches. Facebook has used a “social graph” understand how people are connected. “Knowledge Graph” is Google’s term for how it is building relationships between different people, places and things and report facts about these entities.

Big Change, Subtle Appearance

Earlier this year, the Wall Street Journal wrote about the coming change. At the time, I felt what was described seemed more an extension of things Google had already been doing rather than a dramatic shift. Now having seen it first-hand, I stand corrected. The WSJ had it right. This is indeed a big change in line with other major launches like Search Plus Your World last January and Universal Search in 2007.

Big change, but I don’t think it’ll be a shocking change to most Google users who will begin seeing it over the coming days on, if they’re searching in US English.

Google will still look largely the same as it does now. Knowledge Graph information flows into new units — they have no official name (and I did ask), so I’ll call them “knowledge panels.” These panels appear to the right of Google’s regular results, rather than disrupt those familiar links:

Knowledge panels don’t always appear, only showing up only when Google deems them relevant. But when Google does think they’re relevant, they’re a pretty cool search exploration tool. When the head of Google Search, Amit Singhal, let me play with the new system following his keynote talk at our SMX London show yesterday, I couldn’t help but think of it like a form of StumbleUpon or channel surfing for search.

Fact Surfing

A search for Star Trek brought up a panel that included a reference to Star Trek: Voyager, my favorite of all the series. Jumping to explore that, the Voyager box included a reference to Babylon 5, another favorite sci-fi show of mine. Jumping to that box, there was a reference to Claudia Christian, who wonderfully played one of the main characters in Babylon 5, Susan Ivanova. I surfed over for a look.

If you’ve ever started reading a Wikipedia page and then gotten lost jumping from one topic to another, that’s the experience I think many are about to discover with Google. You’ll not only discover answers to factual questions, but you’ll likely quickly explore more than you had planned and have fun doing it.

3.5 Billion Facts About 500 Million Objects

Google says it has compiled over 3.5 billion facts, which include information about and relationships between 500 million objects or “entities,” as it sometimes calls them. In general, entities are persons, places and things. You know, nouns.

In particular, these are just some of the categories of objects Google has facts about:

  • Actors, Directors, Movies
  • Art Works & Museums
  • Cities & Countries
  • Islands, Lakes, Lighthouses
  • Music Albums & Music Groups
  • Planets & Spacecraft
  • Roller Coasters & Skyscrapers
  • Sports Teams

Again, those are just some of the categories. The relationships are also as important as the facts. The relationships allow the Knowledge Graph to know which actors to list for a particular movie or which spacecraft have visited  a planet.

The Most Popular Facts

How do you keep from getting overwhelmed with useless facts? Google picks out the facts for each object that are most sought in relation to that object.

“We are showing all the things that people look for in a given query,” Singhal told me.

Consider these two knowledge panels, one for Simpson’s creator Matt Groening, the other for architect Frank Lloyd Wright (you can click to enlarge):

For both, you’re told when they were born and where they were educated. After that, the remaining facts shown differ.

Only Groening has facts about his parents and siblings listed. Why? Look closely at the names: Margaret (Marge), Homer, Lisa. Groening named characters after his own family. Looking at searches related to Groening, Google can tell these are commonly sought answers.

For Groening, the books he’s authored are listed. For Wright, his famous buildings are. That makes sense. People are far more interested in structures by Wright than by books by him. Indeed, Google’s autocomplete suggestions — which are based on the most popular terms related to a core search topic — reflect this:

I found it fascinating to see what was shown, as I ran through various classes of searches. For Disneyland, popular rides were shown. For a ride like Space Mountain, the duration was shown (really, only 3 minutes?). For an astronaut, I was shown the missions and overall time they’d spent in space (how cool to have that as a fact about yourself). For Buckingham Palace, the size of floor space was listed. For Larry Page and Mark Zuckerberg, their estimated net worth was shown.

Each knowledge panel has a “People also search for” area at the bottom which lists related people, places or things. Again, the relationships are determined by looking at search data. People who search for Groening, for example, often search for David X. Cohen, who co-created Futurama with Groening.

For search marketers, or anyone interested in how people search, these panels have become another great discovery resource along with keyword research tools like Google Trends, Google Insights, Google Correlate or the AdWords Keyword Tool.

Facts But Not Actions

One thing I found lacking was that the knowledge panels I saw often lacked links to let people take actions related to these objects. For example, one of the popular things people want in relation to Buckingham Palace is to book tickets for tours. However, the panel had no options for this.

In contrast, the new “Snapshots” announced (but still about a week from going live) as part of Bing’s relaunch last week are heavy on trying to help people do things like book tickets or reservations.

Why not have actions?

“We will, of course, explore that, but right now, we just want to take it out and see how it works,” Singhal said.

Occasionally you can take actions via the links to some of the source providers of facts, as with some music searches that might credit Songkick or StubHub.

Which Andromeda Did You Mean?

For some searches, there may be more than one entity that Google has facts for related to a search. In these cases, rather than make the wrong guess, Google will put up a “See results box” as shown below for Andromeda:

Andromeda could mean, in Google’s Knowledge Graph, the galaxy, the TV show or the Swedish band. This box, also known as a disambiguation box, allows people to make the right choice.

Where Do The Facts Come From?

How does Google know any of these facts? Google Squared was an initial attempt in 2009 to extract facts from the web. Google still has that technology, but the service was never that impressive on accuracy and closed as standalone site last year.

Rather, it was Google’s purchase of Metaweb in 2010 that really jump-started the Knowledge Graph. Metaweb was building both the relationships and, though Freebase, a database of facts.

Since that time, Singhal said Google’s massively grown the fact database. Contributions happen with Freebase, but data also comes from publicly-available sources like Wikipedia and The CIA World Factbook and even information out of Google Books. Beyond that, Google also licenses data from others.

“Wherever we can get our hands on structured data, we add it,” Singhal said.

Fixing Bad Data

Drawing from Wikipedia and other public sources means that there’s no guarantee that the facts are accurate. That’s why the knowledge panels on Google all have a “Report a problem” link at the bottom.

If you click on that, you can then indicate if any particular fact is incorrect. Singhal said that Google will use a combination of computer algorithms and human review to decide if a particular fact should be corrected

If Google makes a change, the source provider is told. This mean, in particular, Wikipedia will be informed of any errors. It doesn’t have to change anything, but apparently the service is looking forward to the feedback.

“They really are excited about it. They get to get feedback from a much bigger group of people,” Singhal said.

Will Publisher Traffic Drop?

Search engines have increasingly moved toward showing direct answers in their results over the years. Such efforts have worried some publishers, leaving them wondering if they’ll be left out of receiving search traffic. After all, if search engines provide answers right within their results, why would anyone click away?

Google’s Knowledge Graph is going to massively increase the number of direct answers shown, which will almost certainly renew concerns.

Singhal’s response is that publishers shouldn’t worry. He said that most of these types of queries, Google has found, don’t take traffic away from most sites. Part of this seems to be that the boxes encourage more searching, which in turn still eventually takes people to external sites.

Still, some are going to lose out, he admits. But he sees that as something that was going to happen inevitably, anyway, using a “2+2″ metaphor. If people are searching for 2+2, why shouldn’t Google give a direct answer to that versus sending searchers to a site? By the way, Google does do math like this already and has for years.

Below, you can hear Singhal talk more about this when asked by a member of the audience at SMX London yesterday:

YouTube Preview Image

My concern is what happens if publishers have compiled great information that someone at Wikipedia or Freebase harvests into a database. For example, if a Disneyland fan site has organized a list of ride durations by doing original legwork, what credit do they get if that data is used? Facts can’t be trademarked, at least in the US, so anyone can help themselves assuming they don’t duplicate the exact format or presentation.

Google does list credit links to places like Wikipedia. In turn, Wikipedia does give credit (albeit in a way that doesn’t help search rankings) to the sources it draws from. But that puts actual source material two clicks away from the searcher, assuming the searcher wants to go beyond the fact they already received.

This is one that has to be watched closely. As I wrote before, it seems likely the Knowledge Graph will impact a relatively small set of sites that focus on facts, sites that already likely exposing answers in their listing descriptions and so not getting traffic anyway. But we’ll see.

It’s also important to remember that the “main” results aren’t disappearing. Consider again the Frank Lloyd Wright search, this time with the knowledge panel in context with the regular results:

As you can see, links to sites outside of Google remain to the left and in the most viewed area of a search results page.

Being Included

What if you want to be part of the new knowledge panels and Knowledge Graph in general? Singhal said that at the moment, there’s no mechanism designed for sites to do this. IE, if you run a site about Frank Lloyd Wright, there’s no way to be associated as some type of suggested source for the Frank Lloyd Wright panel.

Potentially, you could head over Freebase, open an account and contribute. Of course, I’m pretty sure adding your blog to a horrible list of blogs like this isn’t going to help. Maybe other categories might be more successful, but I’d hold off, for the moment.

Tagging parts of your pages with commonly used schema might be helpful, though I wouldn’t do this solely in hopes of getting your facts into the Knowledge Graph. The articles below have more about using schema:

Ads, Mobile & Tablet Formats

Anyone familiar with Google’s ads will immediately wonder what happens when the panel shows.

Singhal said that if there are also ads along with a knowledge panel for any search, the ads will still display. Google also has different formats for when a query has a few, many or no ads. I haven’t seen these, but I’ll try to update as they become visible after the launch.

In addition, Google also uses special formats to make the panels work well on tablet and mobile devices, he said. They aren’t restricted to just desktop search, so that’s good news for those of you who want an easier time to cheat at pub and bar quiz nights.

Sadly, there’s no way to just search the Knowledge Graph directly. It only appears with regular Google Search.

The Competition

Google’s not alone in having a knowledge graph, of course. Wolfram Alpha, launched in 2009, has continued to refine its service. It got a big boost being picked as a search partner by Apple to help power Siri (even though that recently embarrassed Apple on a particular search about smart phones).

As for Bing, it has a partnership with Wolfram Alpha plus owns Powerset technology that, somewhat similar to the Knowledge Graph, tries to deeply understand the meanings of words, rather than just really match patterns of letters.

But Bing hasn’t really seemed to capitalize on either its Wolfram partnership nor Powerset. Really, the Knowledge Graph seems to be going more head-to-head with Wolfram Alpha. Does it?

“Wolfram is far more computational,” Singhal said, explaining that Wolfram Alpha’s goal seems to be finding ways that you can effectively use facts in computations.

For example, you can enter cars in california / california population into Wolfram Alpha to have it take those two facts and come up with an average (about 1 car for every two people, by the way, using 2009 data).

Google’s not trying to perform these types of calculations. The focus is instead on providing popular facts.

The Future

The big picture, of course, is that some day the Knowledge Graph won’t just be used for facts. Instead, if Google can better tag actual web pages to entities, then it can better understand what those pages are about and related to, which might increase the relevancy of its regular results.

That’s down the line, as are many other changes to the knowledge panel themselves. Today represents only a start.

“This is just a baby step, in my view, to expose this to our users,” Singhal said.

To learn more about the Google Knowledge Graph, see coverage from others across the web organized here on Techmeme, the official Google blog post, plus the official video, below:

YouTube Preview Image

Related Articles

Related Topics: Channel: SEO | Features: Analysis | Google: Knowledge Graph | Google: Web Search | Search Features: Direct Answers | Top News | Wolfram Alpha


About The Author: is a Founding Editor of Search Engine Land. He’s a widely cited authority on search engines and search marketing issues who has covered the space since 1996. Danny also serves as Chief Content Officer for Third Door Media, which publishes Search Engine Land and produces the SMX: Search Marketing Expo conference series. He has a personal blog called Daggle (and keeps his disclosures page there). He can be found on Facebook, Google + and microblogs on Twitter as @dannysullivan.

Connect with the author via: Email | Twitter | Google+ | LinkedIn


Get all the top search stories emailed daily!  


Other ways to share:

Read before commenting! We welcome constructive comments and allow any that meet our common sense criteria. This means being respectful and polite to others. It means providing helpful information that contributes to a story or discussion. It means leaving links only that substantially add further to a discussion. Comments using foul language, being disrespectful to others or otherwise violating what we believe are common sense standards of discussion will be deleted. Comments may also be removed if they are posted from anonymous accounts. You can read more about our comments policy here.
  • Karthik kumar

    After a long time, as a searcher, I found myself exploring information using Google (and not wanting to stop), just like when I first started using Wikipedia (for an older equivalent, the Oxford thesaurus book.. :D). Well, this is a nice experience.

  • Kaushalam

    For general awareness this is an essential resource as instantly one can find about people, places or thing he want to look more information. I things this could be a great resources for students to increase their general knowledge base.

  • Webstats Art

    Daveintheuk has given an intelligent reply. Google want to use everyone’s work for their own benefit. Google search is basically an accepted scraper making it’s own content farms on it’s own pages. Since it is not illegal to embed other people’s work in Google search (not yet anyway) then why bother visiting someone’s website that could of taken years to build. 

  • Webstats Art

    A lot of legitimate businesses providing excellent services have been damaged since early 2011 with the Google Panda etc etc. Fortunately many in the business world are now waking up and viewing google as the enemy because they see Google as a middle man in their business. Google get in the way of direct contact with clients through their search engine. They say that Google juggle with people’s success depending on some algorithm that makes certain sites rule and other sites fail. So it does not matter if you are making stuff out of a garage or spending millions to produce an excellent product/service because at the end of the day you are stuck with google who decide your fate.

    Although the above comments may seem negative it has actually made people think deeply about the irrelevancy of a search engine for their business and the need for those businesses to find more direct ways to contacting their clients without Google and once they have the clients, they block google from what is going on. A good example of a very clever company using this technique is Facebook. I don’t really use Facebook but I respect the people behind the technology because they know what is going on and they have done well in protecting themselves.

  • Katherine

    my neighbor’s aunt makes $61 hourly on the computer. She has been unemployed for nine months but last month her income was $13180 just working on the computer for a few hours. Go to this web site and read more CashLazy.c&#111m

  • Bryce Hanson

    Sounds like they are trying to emulate Khan Academy’s Knowledge map.

  • MarketingHumenews

    I think we are caught between what we want out of a search engine as individuals and what we want out of a search engine as website owners. As an individual looking for information the quicker and easier I can get it the better. As a website owner, I want people to get it quickly also, but I would prefer they take that extra time to click my site to get it. Realistically though it is better for the searcher to not have to take that extra click so I can understand Google refining the search.

  • SargentManuela

    my friend’s aunt made $17398 the previous week. she is making income on the internet and bought a $578000 house. All she did was get lucky and try the steps written on this website===>> ⇛⇛⇛⇛►

  • SalinasMorris

    my buddy’s mother go t paid $21508 the previous week. she is making money on the internet and bought a $386500 house. All she did was get fortunate and put into work the instructions explained on this web site===>> ⇛⇛⇛⇛►

  • AgentsOfValue

    Knowledge Graph will be useful for everyone, especially for those curious minds who never stop to get hungry with facts. This is going to be a useful and helpful feature to all researchers, writers and bloggers.

  • Phoebe

    my neighbor’s aunt makes $73 an hour on the laptop. She has been fired from work for 7 months but last month her paycheck was $20065 just working on the laptop for a few hours. Read more on this web site CashLazy.c&#111m

  • Robyn

    my best friend’s sister-in-law makes $74 hourly on the computer. She has been without work for 7 months but last month her check was $21606 just working on the computer for a few hours. Read more here CashLazy.c&#111m  

  • NguyenMarquita84

    my friend’s aunt made $17398 the previous week. she is making income on the internet and bought a $578000 house. All she did was get lucky and try the steps written on this website===>> ⇛⇛⇛⇛►

  • Webstats Art

    The knowledge graph has changed since you wrote this article. Google are using it to promote Google+ by putting links to googleplus individual profiles underneath the pictures of people without their permission.

  • Jeff Kean

    Kirk is a bum, Picard all the way!!!

Get Our News, Everywhere!

Daily Email:

Follow Search Engine Land on Twitter @sengineland Like Search Engine Land on Facebook Follow Search Engine Land on Google+ Get the Search Engine Land Feed Connect with Search Engine Land on LinkedIn Check out our Tumblr! See us on Pinterest


Click to watch SMX conference video

Join us at one of our SMX or MarTech events:

United States


Australia & China

Learn more about: SMX | MarTech

Free Daily Search News Recap!

SearchCap is a once-per-day newsletter update - sign up below and get the news delivered to you!



Search Engine Land Periodic Table of SEO Success Factors

Get Your Copy
Read The Full SEO Guide