Wednesday, October 31, 2012

Large Scale Language Modeling in Automatic Speech Recognition



At Google, we’re able to use the large amounts of data made available by the Web’s fast growth. Two such data sources are the anonymized queries on google.com and the web itself. They help improve automatic speech recognition through large language models: Voice Search makes use of the former, whereas YouTube speech transcription benefits significantly from the latter.

The language model is the component of a speech recognizer that assigns a probability to the next word in a sentence given the previous ones. As an example, if the previous words are “new york”, the model would assign a higher probability to “pizza” than say “granola”. The n-gram approach to language modeling (predicting the next word based on the previous n-1 words) is particularly well-suited to such large amounts of data: it scales gracefully, and the non-parametric nature of the model allows it to grow with more data. For example, on Voice Search we were able to train and evaluate 5-gram language models consisting of 12 billion n-grams, built using large vocabularies (1 million words), and trained on as many as 230 billion words.


The computational effort pays off, as highlighted by the plot above: both word error rate (a measure of speech recognition accuracy) and search error rate (a metric we use to evaluate the output of the speech recognition system when used in a search engine) decrease significantly with larger language models.

A more detailed summary of results on Voice Search and a few YouTube speech transcription tasks (authors: Ciprian Chelba, Dan Bikel, Maria Shugrina, Patrick Nguyen, Shankar Kumar) presents our results when increasing both the amount of training data, and the size of the language model estimated from such data. Depending on the task, availability and amount of training data used, as well as language model size and the performance of the underlying speech recognizer, we observe reductions in word error rate between 6% and 10% relative, for systems on a wide range of operating points.

Celebrating Leonardo Torres-Quevedo, Spanish inventor of the world’s first computer game

What does a cable car in Niagara Falls have to do with the world’s first chess-playing machine? Surprisingly, both were inventions of Spanish civil engineer Leonardo Torres-Quevedo. Next week, as part of our ongoing effort to celebrate Europe’s computing heritage, we’re commemorating Torres-Quevedo’s legacy and his remarkable machine—"El Ajedrecista" (in English, "The Chess Player")—in partnership with the Telecommunication Engineering department of the Technical University of Madrid.

Photo thanks to Wikimedia Commons

Torres-Quevedo’s inventions span many fields. He was the second in the world to demonstrate wireless remote control, beaten to the post only by Nikola Tesla. His designs for airships were used by both the French and British during WWI. He was a global leader in cable car design, creating the “Spanish aero car” over the Niagara Whirlpool which, nearly a century on, remains a tourist attraction. However, his most remarkable achievements were in the field of automation, developing machines that are antecedents to what we now call computers and robots.

Torres-Quevedo’s ambitions were bold. As Scientific American proclaimed in 1915: “He would substitute machinery for the human mind.” In the 1890s, Torres-Quevedo built a series of mechanical devices that solved algebraic equations. In 1920 he wowed a Paris audience with an electromechanical arithmometer with a typewriter attachment. You simply typed a formula—say, “24x48”—and the machine would calculate and automatically type the answer “=1152” in reply.

But El Ajedrecista, an algorithmically powered machine that could play an end-game of chess against a human opponent completely automatically, is his most notable creation. Although it’s a far cry from Deep Blue, El Ajedrecista can lay claim to being the world’s first (analog) computer game.

.
Photos thanks to Museo Torres Quevedo

The machine didn’t just calculate its moves—it had mechanical arms that physically moved its pieces, in the form of electrical jacks, across a grid. In later models the arm mechanism was replaced by magnets, and play took place on a more ordinary-looking chess board. You couldn’t cheat the machine as it could spot illegal moves; and you couldn’t win, as the game always started at a point (machine’s King and Rook versus human’s King) from which the machine could never lose.

In honor of El Ajedrecista’s 100th birthday, we’re working with the Telecommunication Engineering department of the Technical University of Madrid to stage a conference commemorating Torres-Quevedo’s legacy. The conference, taking place on November 7, will feature lectures and panel discussions, as well as an exhibition of Torres-Quevedo’s devices—including El Ajedrecista itself. Attendance is free—if you want to join us, request an invitation.

Google’s most advanced voice search has arrived on iOS

When you have a question, finding the answer should be effortless—wherever you are and whatever device you’re using. The new Google Search app for iPhone and iPad helps you to do just that with enhanced voice search that answers any question with the comprehensive Google search results you know and love.

Fast and accurate voice recognition technology enables Google to understand exactly what you’re saying. Getting an answer is as simple as tapping on the microphone icon and asking a question like, “Is United Airlines flight 318 on time?” Your words appear as you speak, you get your answer immediately and—if it’s short and quick, like the status and departure time of your flight—Google tells you the answer aloud.

You can get answers to an increasingly wide variety of questions thanks to Knowledge Graph, which gives our search technology an understanding of people, places and things in the real world. Here are a few of the questions that Google can answer:
  • “What does Yankee Stadium look like?” Google will show you hundreds of pictures instantly.
  • “Play me a trailer of the upcoming James Bond movie.” The trailer starts playing immediately right within Google Search. 
  • “When does daylight savings time end?” The answer will appear above the search results, so you can set your clock without having to click on a link. 
  •  “Who’s in the cast of The Office?” See a complete cast list and find out who made you crack up last night. 


Download the Google Search App on your iPhone, iPad or iPod Touch and find out how much wood a woodchuck would chuck (if a woodchuck could chuck wood).

Tuesday, October 30, 2012

More Art Project online for you to explore

Back in April we announced a major expansion of the Google Art Project. Since then 15 million people have explored the paintings, sculptures, street art and photographs contributed by our partners. From today the number of treasures you can view is increasing by more than 10% as 29 new art organizations from 14 countries bring their collections online.

A wide range of global institutions, large and small, well-known and less traditional, are represented. Explore contemporary works at the Istanbul Modern Art Museum, admire works from the Art Gallery of South Australia (who have contributed almost 600 objects) and access the treasures of the famous Museum of Palazzo Vecchio in Italy and Princeton University. This round has also seen contributions from more unusual sources including a collection from the National Ballet of Canada, pre-Columbian art from Peru and decorative arts from China.


Now that the total number of objects online is more than 35,000, we've turned our attention towards thinking of different ways for you to experience the collections.

The first is a great educational tool for art students, enthusiasts or those who are simply curious. A “Compare” button has been added to the toolbar on the left of each painting. This allows you to examine two pieces of artwork side-by-side to look at how an artist’s style evolved over time, connect trends across cultures or delve deeply into two parts of the same work. Here's an example: place an early sketch of Winslow Homer's 'The Life Line' from the Cooper-Hewitt, National Design Museum next to the completed painting from the Philadelphia Museum of Art. Comparing them in this way allows you to see how the artist's vision altered (or not) over the life of the work.


Beyond following us and discussing great art on our Google+ page, we have also created a Hangout app within the Art Project so that you can share your favorite collections and perhaps give your friends a personal guided tour. If there is a budding museum guide or an art critic within any of you it can finally be unleashed! Watch this video to see how it works.



Around 180 partners have contributed their works to the Art Project so far, more than 300,000 of you have created your own online galleries and we've had more than 15 million visitors since our last launch in April. The cultural community has invested great time and effort to bring these masterpieces online. Watch this space for more to come.

Public Alerts on Google Search and Maps for Android for superstorm Sandy preparedness information

Earlier today we posted about efforts to provide information to those affected by the former hurricane and now superstorm Sandy.

We also want to let you know that Public Alerts are now available on Google Search & Maps in your browser, on Google Maps for Android and also on Google Now for Android devices running Jellybean.

Public Alerts provide warnings for natural disasters and emergency situations. They appear based on targeted Google searches, such as [Superstorm Sandy], or with location-based search queries like [New York]. In addition to the alert, you’ll also see relevant response information, such as evacuation routes, crisis maps or shelter locations.

We were planning on announcing the new features in a few days, but wanted to get them out as soon as possible so they can be helpful to people during this time.

This is part of our continuing mission to bring emergency information to people when and where it is relevant. Public Alerts are primarily available in English for the U.S., but we are working with data providers across the world to expand their reach.

If you are searching for superstorm Sandy, you’ll see content at the top of the Search page specific to this crisis. For other searches, you’ll see public alerts where and when they are live.


Public Alerts on desktop search

Public Alerts on mobile

Desktop search showing content for Sandy-related query

We’re able to gather relevant emergency safety information thanks to a strong network of partners, including NOAA and USGS. Their commitment to open standards like the Common Alerting Protocol (CAP) is what makes this all possible. We’ve also developed partnerships to bring you even more relevant alerts in the future, including local emergency data from Nixle.

To learn more about Public Alerts, visit our Public Alerts homepage. If you’re a data provider, and would like to contribute to our efforts, please see our FAQ.

We hope that this information makes it easier for you to stay safe.

Get Ready to Vote with Google

Every four years in the United States, people prepare to head to the polls and increasingly search for information about how to register to vote, where to vote and who is on their ballot. Even though it is 2012, important voting information is disorganized and hard to find on the Internet. To help voters research candidates and successfully cast their ballot on Election Day, we’ve launched our new Voter Information Tool.


You can enter your address to find information on your polling place, early vote locations, ballot information with links to candidates’ social media sites and voting rules and requirements. The tool is easy to embed on any website and is open source so developers can modify it to create custom versions. We're working with a number of media partners to ensure the tool is accessible across the web, and partners like Foursquare and AT&T are doing great work building apps on our Civic Information API.

We hope this tool will help make getting to the polls and casting your ballot as simple as possible.



(Cross-posted on the Politics and Elections blog)

Monday, October 29, 2012

New Crisis Response maps feature preparedness information for Hurricane Sandy

Already responsible for a reported 41 deaths across the Caribbean, late-season Hurricane Sandy is expected to make landfall again early this week on the East Coast of the United States.

Some are calling the hurricane “Frankenstorm” due to its potential mix of both winter and tropical cyclone weather. Regardless of what you call it, we hope that you get the information you need to make preparations and stay safe if you are in the area. It has the potential to be one of the worst storms the area has seen in decades.

The Google Crisis Response team has assembled a Hurricane Sandy map to help you track the storm’s progress and provide updated emergency information.

 

On the map, you’ll find the following emergency preparedness information:
We’ve also launched a map specific to New York City, featuring evacuation zone information from NYC Open Data, open shelters, weather information and live webcams.


You can easily share and embed these maps on your website — just hit the “Share” button at the top of the map to get the HTML code. We’ll continue to update these maps as more information becomes available.

Posted by Ka-Ping Yee, Software Engineer, Google Crisis Response

(Cross-posted on the Google.org blog)

Nexus: The best of Google, now in three sizes

People increasingly have more than one device, and they switch between them many times a day. Nexus—Google’s hardware line for Android devices—gets rid of the hassle. Just sign in with your Google Account and everything is there ready to go, whatever device you’re using: photos, emails, contacts, bookmarks, even your entertainment on Google Play.

Today, we’re excited to announce three great new Nexus devices … in small, medium and large. And they all run Android 4.2, a new flavor of Jelly Bean—which includes the latest version of Google Now and other great new features.




Nexus 4 with Google Now and Photo Sphere 
Nexus 4 is our latest smartphone, developed together with LG. It has a quad-core processor which means it's super fast, a crisp 4.7" (320 ppi) display that's perfect for looking at photos and watching YouTube, and with wireless charging you just set the phone down on a charging surface to power it up, no wires needed. While Nexus 4 is incredibly powerful under the hood, it also features the latest version of Jelly Bean, Android 4.2—the simplest and smartest version of Android yet. Starting with the camera, we've reinvented the photo experience with Photo Sphere, which lets you capture images that are literally larger than life. Snap shots up, down and in every direction to create stunning 360-degree immersive experiences that you can share on Google+ with friends and family—or you can add your Photo Sphere to Google Maps for the world to see.

Android 4.2 brings other great goodies like Gesture Typing, which lets you glide your finger over the letters you want to type on the keyboard—it makes typing fast, fun and a whole lot simpler. Android 4.2 also adds support for wireless display so you can wirelessly watch movies, YouTube videos and play games right on your Miracast-compatible HDTV.

Learn more about all of the new features of Android 4.2, Jelly Bean, here.

Google Now—even more useful 
We designed Google Now to make life simpler by giving you the right information at just the right time in easy to read cards, before you even ask. And the feedback has been awesome. So today we’re adding more cards that we hope you’ll find useful. Flight information, restaurant reservations, hotel confirmations and shipping details—how often have you found yourself wading through your email to get this information at the last moment? So next time you book a table for dinner, you’ll get a reminder with all the details without ever having to lift a finger. You’ll also get cards for nearby attractions, interesting photo spots, movies times at nearby theaters or concerts by your favorite artists.

Nexus 7: Thin, light and now even more portable
Nexus 7 brings you the best of Google–YouTube, Chrome, Gmail, Maps–and all the great content from Google Play in a slim, portable package that fits perfectly in your hand. To give you more room for all that great content you can now get Nexus 7 with 16GB ($199) or 32GB ($249) of storage. But we also wanted to make this highly portable tablet even more mobile. So we added HSPA+ mobile data. Nexus 7 is now also available with 32GB and HSPA+ mobile ($299), which can operate on more than 200 GSM providers worldwide, including AT&T and T-Mobile in the U.S. 

Nexus 10: Powerful and shareable 
Nexus 10 is the ultimate tablet for watching movies or reading magazines. We wanted to build a premium entertainment device, so we partnered with Samsung to do just that. Nexus 10 is the highest resolution tablet on the planet with a 10.055" display at 2560-by-1600 (300ppi), that's over 4 million pixels right in your hands. It comes with a powerful battery that will get you up to nine hours of video playback and more than 500 hours of standby time. With a set of front-facing stereo speakers, you can watch movies right from your Nexus 10 and they simply sound awesome. But what makes Nexus 10 unique is that it's the first truly shareable tablet. With Android 4.2, you can add multiple users and switch between them instantly right from the lockscreen. We believe that everyone should have quick and easy access to their own stuff -- email, apps, bookmarks, and more. That way, everyone can have their own home screens, their own music, and even their own high scores.

Google Play: More entertainment, more countries
We’ve recently added a ton of great new entertainment to Google Play, such as movies and TV shows from Twentieth Century Fox. Earlier this year we expanded our service beyond movie rentals and now you can purchase movies and build a library of your favorites in Google Play. Today we’re bringing movie purchasing to more countries - Canada, the U.K., France, Spain and Australia.

We’re also excited to announce two new partnerships. We’re now working with Time, Inc. to bring you even more magazines like InStyle, PEOPLE, TIME and others. And we’ve partnered with Warner Music Group who will be adding their full music catalog with new songs coming each day. We’re now working with all of the major record labels globally, and all the major U.S. magazine publishers, as well as many independent labels, artists and publishers.

On November 13, we're bringing music on Google Play to Europe.  Those of you in the U.K, France, Germany, Italy and Spain will be able to purchase music from the Google Play store and add up to 20,000 songs—for free—from your existing collection to the cloud for streaming to your Android devices or web browser. We’re also launching our new matching feature to streamline the process of uploading your personal music to Google Play. We’ll scan your music collection and any song we match against the Google Play catalog will be automatically added to your online library without needing to upload it, saving you time. This will be available in Europe at launch on November 13 and is coming to the U.S. soon after. This will all be for free—free storage of your music, free matching, free syncing across your devices and free listening.

Great value
We’ve always focused on building great devices at great value.  And we think today’s devices offer the very best that money can buy. Here are more details on when and where you can pick up your next Nexus device:
  • Nexus 4: 8GB for $299; 16GB for $349; available unlocked and without a contract on 11/13 on the Google Play store in the U.S., U.K., Australia, France, Germany, Spain and Canada. The 16GB version will also be available through T-Mobile for $199, with a 2-year contract (check here for more details).
  • Nexus 7: 16GB for $199 and 32GB for $249; available in the U.S., U.K., Australia, France, Germany, Spain, Canada and Japan, and also through our retail partners Gamestop, Office Depot, Office Max, Staples and Walmart.
  • Nexus 10: 16GB for $399; 32GB for $499; available on 11/13 in the Google Play Store in the U.S., U.K., Australia, France, Germany, Spain, Canada and Japan. You'll also be able to purchase the 32GB version in more than 2,000 Walmart stores in the U.S.
A Nexus device is much more than simply a phone or tablet. It’s your connection to the best of Google—all of your stuff and entertainment, everywhere you go with no hassle.  Now you have three new Nexus devices, a new improved version of Jelly Bean and more entertainment than ever before—all available on Google Play. The playground is open.


with the availability of Nexus 10 at Walmart.

Friday, October 26, 2012

Get your Google back

If you’ve moved to Windows 8 and are getting acquainted with it, you may be looking for a couple of your favorite Google products that you use every day. To help you get the best experience possible on Google and across the web, we’ve designed and built a new Google Search app and Chrome browser for Windows 8 and created a simple site to help you get your Google back.

 

The Google Search app comes with a clean and recognizable user interface. Our new voice search lets you naturally speak questions. The image search and image previews are built for swiping. And, as usual, you get immediate results as you type with Google Instant. The doodles you enjoy on special occasions will be right there on the homepage and even show up on the Google tile on your start screen.


The Chrome browser is the same Chrome you know and love, with some customizations to optimize for touchscreens, including larger buttons and the ability to keep Chrome open next to your other favorite apps. It delivers the fast, secure web experience you’ve come to expect from Chrome on all your devices.


To get both Google Search and Chrome installed on your Windows 8 machine, head to our site and learn how to get your familiar Google apps back.

Wednesday, October 24, 2012

Trekking the Grand Canyon for Google Maps

In our ongoing effort to create the perfect map—one that’s as comprehensive, accurate and easy to use as possible—we’ve gone well beyond just the streets. Through the Street View feature on Google Maps, you’ve been able to explore panoramic views of amazing places around the world ranging from the Swiss Alps to the Amazon to Antarctica, and a variety of urban cities, university campuses, ancient ruins and ski resorts as well.

Today, demonstrating the rocky and rugged paths we’ll travel to make Google Maps even more complete, we’re collecting imagery from a place no car, trike or snowmobile has ever been before. On its first official outing, the Street View team is using the Trekker—a wearable backpack with a camera system on top—to traverse the Grand Canyon and capture 360-degree images of one of the most breathtaking natural landscapes on the planet.

Operations Manager Steve Silverman (left) and Product Manager Ryan Falor (right) hiking the Bright Angel Trail on the South Rim with Trekkers 

The narrow ridges and steep, exposed trails of the Grand Canyon provide the perfect terrain for our newest camera system. The Trekker—which its operator controls via an Android phone and automatically gathers photos as he walks—enables the collection of high-quality imagery from places that are only accessible on foot.

Falor controlling the the Trekker with his Android device 

This week, photos are being gathered from portions of the South Rim at Grand Canyon National Park, including the ridge, the famous Bright Angel Trail, South Kaibab Trail and more. These panoramic views will soon be live on Google Maps, giving everyone from real-life visitors to armchair travelers the opportunity to marvel at this beautiful, majestic site from the comfort of their computers or mobile devices.

The team hiking the Bright Angel Trail 

So get ready for the virtual adventure that awaits! And in the meantime, we’ll keep on trekken’ and working hard to bring you panoramic imagery of more visually stunning places we have yet to explore and share on Google Maps.

Watch the Hajj from the Grand Mosque of Mecca live on YouTube

Tomorrow marks the start of the observance of Eid El Adha, celebrated by the world’s 1.5 billion Muslims. As part of this holiday, nearly 2.5 million Muslims will participate in the world’s largest pilgrimage to perform the ritual of Hajj. This year, millions more from around the world will be able to experience the ritual via the live stream from Mecca, Saudi Arabia on the Saudi Ministry of Culture and Information's YouTube channel.



The Hajj represents one of the five pillars of Islam; it requires all Muslims around the world who are able-bodied and can afford it to perform the pilgrimage once in their lifetime. The pilgrimage occurs from the 8th to 12th day of Dhu al-Hijjah, the 12th and last month of the Islamic calendar.

This live stream was made possible by our cooperation with the Saudi Ministry of Culture and Information, which also broadcast Islamic prayers live from the Grand Mosque in Mecca during this year’s Ramadan.

Watch on www.youtube.com/makkahlive.

Posted by Abdulrahman Tarabzouni, Head of Emerging Arabia, Google

Thursday, October 18, 2012

The new Chromebook, for everyone

As a kid growing up in India, I was fascinated with computers and the endless possibilities they presented. I had to wait until college to finally get my hands on one in the computer lab and since then began dreaming of a world where everyone could have access to one. We’re not quite there yet, but every day we get a bit closer.

A few years ago, we set out on a journey to build a better computer that’s faster, simpler and more secure. When we introduced a few Chromebooks into the market, many of you early adopters joined us on this journey. For folks living entirely in the cloud, the Chromebook is now a primary computer.

Many people use the Chromebook today as the perfect additional computer for their home. For families, it’s easy to use and share: for kids doing homework on the couch, parents catching up on emails at the kitchen counter and grandparents staying connected on video chat. There’s no need to worry about security updates and maintenance is easy; all you need to do is charge the battery. It just works.

This gets to the heart of the Chromebook vision. In order to have one, two or more computers around the house, they need to be easy to use and much more affordable. So together with Samsung, we designed a new laptop—the new Samsung Chromebook for $249—the computer for everyone.

    

The new Chromebook is a great computer at any price, but it’s an incredible computer at $249. It’s one of the lightest laptops on the market. You can easily carry it around all day—it’s 2.5 pounds, a mere 0.8 inches thick, with more than 6 hours of battery life for the typical user. And with 100 GB of free storage on Google Drive*, you can get to all of your stuff anytime, anywhere.

Even with its compact design, it’s packed with performance—it boots up in less than 10 seconds and resumes instantly. High-resolution videos (in 1080p) are beautiful to watch and when using the touchpad, you’ll notice smooth scrolling due to a hardware-accelerated user interface. And as you‘d expect from a Chromebook, it’s easy to share with others. Everyone—mom, dad, grandparents, tech lovers, tech haters—can have separate accounts where all of their stuff is kept safe. Finally, if you’re an active Google user of products like Gmail, Drive, Search, Maps, YouTube, Play or Google+ Hangouts, everything just works seamlessly.

Starting today, the new Samsung Chromebook is available for pre-order online from Amazon, Best Buy, PC World and other retailers. Next week it will be available for sale from these same online retailers as well as the Google Play store. You can also buy them at over 500 Best Buy stores across the U.S and over 30 PC World and Currys stores in the U.K.

So if you ever felt it was too complicated and too expensive to have an additional computer (or two), we hope you (and the entire family) will give the new Chromebook a try.

Posted by Sundar Pichai, SVP, Chrome & Apps

*You will have 100 GB of free storage for 2 years, starting on the date you redeem the offer on eligible Chrome devices. 

Ngram Viewer 2.0



Since launching the Google Books Ngram Viewer, we’ve been overjoyed by the public reception. Co-creator Will Brockman and I hoped that the ability to track the usage of phrases across time would be of interest to professional linguists, historians, and bibliophiles. What we didn’t expect was its popularity among casual users. Since the launch in 2010, the Ngram Viewer has been used about 50 times every minute to explore how phrases have been used in books spanning the centuries. That’s over 45 million graphs created, each one a glimpse into the history of the written word. For instance, comparing flapper, hippie, and yuppie, you can see when each word peaked:

Meanwhile, Google Books reached a milestone, having scanned 20 million books. That’s approximately one-seventh of all the books published since Gutenberg invented the printing press. We’ve updated the Ngram Viewer datasets to include a lot of those new books we’ve scanned, as well as improvements our engineers made in OCR and in hammering out inconsistencies between library and publisher metadata. (We’ve kept the old dataset around for scientists pursuing empirical, replicable language experiments such as the ones Jean-Baptiste Michel and Erez Lieberman Aiden conducted for our Science paper.)

At Google, we’re also trying to understand the meaning behind what people write, and to do that it helps to understand grammar. Last summer Slav Petrov of Google’s Natural Language Processing group and his intern Yuri Lin (who’s since joined Google full-time) built a system that identified parts of speech—nouns, adverbs, conjunctions and so forth—for all of the words in the millions of Ngram Viewer books. Now, for instance, you can compare the verb and noun forms of “cheer” to see how the frequencies have converged over time:
Some users requested the ability to combine Ngrams, and Googler Matthew Gray generalized that notion into what we’re calling Ngram compositions: the ability to add, subtract, multiply, and divide Ngram counts. For instance, you can see how “record player” rose at the expense of “Victrola”:
Our info page explains all the details about this curious notion of treating phrases like components of a mathematical expression. We’re guessing they’ll only be of interest to lexicographers, but then again that’s what we thought about Ngram Viewer 1.0.

Oh, and we added Italian too, supplementing our current languages: English, Chinese, Spanish, French, German, Hebrew, and Russian. Buon divertimento!

Wednesday, October 17, 2012

Google’s data centers: an inside look

Very few people have stepped inside Google’s data centers, and for good reason: our first priority is the privacy and security of your data, and we go to great lengths to protect it, keeping our sites under close guard. While we’ve shared many of our designs and best practices, and we’ve been publishing our efficiency data since 2008, only a small set of employees have access to the server floor itself.

Today, for the first time, you can see inside our data centers and pay them a virtual visit. On Where the Internet lives, our new site featuring beautiful photographs by Connie Zhou, you’ll get a never-before-seen look at the technology, the people and the places that keep Google running.




  

 In addition, you can now explore our Lenoir, NC data center at your own pace in Street View. Walk in the front door, head up the stairs, turn right at the ping-pong table and head down the hall to the data center floor. Or take a stroll around the exterior of the facility to see our energy-efficient cooling infrastructure. You can also watch a video tour to learn more about what you're viewing in Street View and see some of our equipment in action.



Finally, we invited author and WIRED reporter Steven Levy to talk to the architects of our infrastructure and get an unprecedented look at its inner workings. His new story is an exploration of the history and evolution of our infrastructure, with a first-time-ever report from the floor of a Google data center.

Fourteen years ago, back when Google was a student research project, Larry and Sergey powered their new search engine using a few cheap, off-the-shelf servers stacked in creative ways. We’ve grown a bit since then, and we hope you enjoy this glimpse at what we’ve built. In the coming days we’ll share a series of posts on the Google Green Blog that explore some of the photographs in more detail, so stay tuned for more!

Tuesday, October 16, 2012

The meaning of open, redux

Nearly three years ago I wrote a blog post, "The Meaning of Open," about Google’s commitment to openness and how it makes the entire company better. After it was published I received several thoughtful responses from readers—professors and writers appreciative of the look inside Google, business leaders telling me how open affects their business, grad school students surprised that this was the very opposite of the lock-in strategy they were being taught.

This week we released the "Open" issue of Think Quarterly, which includes a few of my thoughts on how the world of open has progressed in the brief time since that December 2009 post. In short, the impact of open systems has been greater than even my most optimistic self would have predicted back then. Open ecosystems are rewriting the rules—not only in the tech industry but in education, healthcare and governance.

My article, "The Future is Open," is just one of several good reads in the "Open" issue. I hope you enjoy it!

Celebrating Hispanic Heritage Month

September 15 marked the beginning of Hispanic Heritage Month and the start of our third year celebrating the Hispanic community through events and community outreach initiatives. Googlers from our Corporate Social Responsibility Team, Diversity & Inclusion Team, Engineering Industry Team, the Hispanic Googler Network (HGN), and our Community Partners worked together to host 20+ events focused on this year’s theme of Latinos in Technology.

We kicked things off at the U.S. Hispanic Chamber of Commerce (USHCC) National Conference, where two members from our Google Accelerate team worked one-on-one with business owners during matchmaking sessions to consult on the best use of Google tools for their enterprises. Googler Eliana Murillo spoke on a panel titled “Beyond Social Media: The Potential of Technology & the Internet in a Global Economy,” where she shared how tools like Google Analytics, YouTube and Google for Nonprofits can be useful for businesses.

In early October, we ran a Hispanic Heritage Month 2012 Hangout on Air on the Life at Google page with the Latino Community Foundation (LCF). Raquel Donoso (CEO of LCF) and Googlers Hector Mujica (HGN member) and myself shared the history of the partnership and what our respective goals are. They also talked about the Family Health Day at Google & Olympic Games event, which we held at our Mountain View, Calif. headquarters that same week. Health is a pressing issue (PDF) in the Hispanic community; at this event, part of the Binational Health Week, we encouraged guests to have healthier lifestyles by teaching them some easy exercises, how to be active and eat healthy. More than 380+ community members and 50+ Googlers attended.

Last week we wrapped up a series of networking events in partnership with the Society of Hispanic Engineers (SHPE), where more than 400+ technical professionals came to our Seattle, Cambridge, Chicago, Los Angeles, Austin, New York, and Mountain View Offices to network and learn about how Google is supporting the local hispanic technical community.



Finally, today the Hispanic Googler Network is hosting the Bay Area Latino Employee Resource Group (ERG) Networking Reception in Mountain View. The Honorable Aida Alvarez, Chair of the Latino Community Foundation of the Bay Area, will speak to 300+ guests from local Hispanic ERGs in the Bay Area about what LCF is doing to build a better future for Latino children, youth and families in the Hispanic community.

Though the month officially comes to an end today, we’ll continue to support the Hispanic community as a lead sponsor in the LATISM '12 conference, taking place in two weeks. LATISM ‘12 connects Latinos in social media, technology, education, business and health fields to increase their online footprint through the web and Google's tools for small businesses and communities. We’re also participating in the Society of Hispanic Professional Engineers Conference and will soon open up applications for our Hispanic College Fund Google scholarship.

We’ve had a great time celebrating Hispanic Heritage Month, and are already looking forward to next year’s events. We invite you to view the recaps, photos and hangouts on our Life at Google page on Google+ and to visit our Diversity & Inclusion site where you can see more of what we do.

Wednesday, October 10, 2012

Bringing history to life

Today you can discover 42 new online historical exhibitions telling the stories behind major events of the last century, including Apartheid, D-Day and the Holocaust. The stories have been put together by 17 partners including museums and cultural foundations who have drawn on their archives of letters, manuscripts, first-hand video testimonials and much more. Much of the material is very moving—and some is on the Internet for the first time.



Each exhibition features a narrative which links the archive material together to unlock the different perspectives, nuances and tales behind these events. Among others you’ll see:
  • Tragic love at Auschwitz - the story of Edek & Mala, a couple in love who try to escape Auschwitz
  • Jan Karski, Humanity’s hero - first-hand video testimony from the man who attempted to inform the world about the existence of the Holocaust
  • Faith in the Human Spirit is not Lost - tracing the history of Yad Vashem’s efforts to honor courageous individuals who attempted to rescue Jews during the Holocaust
  • Steve Biko - a 15-year-old’s political awakening in the midst of the Apartheid movement featuring nine documents never released in the public domain before
  • D-Day - details of the famous landings including color photographs, personal letters and the D-Day order itself from Admiral Ramsay
  • The Coronation of Queen Elizabeth II - an account of the 1953 Coronation including color photographs
  • Years of the Dolce Vita - a look at the era of the “good life” in Italy including the fashion, food, cars and culture
As with the other archives that we’ve helped bring onto the Internet, including the Dead Sea Scrolls, you can zoom in to see photos in great detail and search through millions of items for a specific country, person, event or date. Watch our video for some guidance about how to find your way around the exhibitions.

The historical collections are the latest chapter in the work of the Google Cultural Institute, following the Art Project, World Wonders and the Nelson Mandela archives. We’re working closely with museums, foundations and other archives around the world to make more cultural and historical material accessible online and by doing so preserve it for future generations.

You can explore the many exhibitions at www.google.com/culturalinstitute. You can also follow us on our Google+ page. What you see today is just the start, so if you’re a partner interested in contributing your own exhibitions, please fill out this form.

Tuesday, October 9, 2012

Communicate more easily across languages in Gmail

Finding the right words can be difficult, especially across languages, and once you choose them, finding a way to type them can be even harder. Try emailing family in Germany, chatting with friends in China or adding a Russian business partner's name to your contacts and you may find yourself limited by the language of your keyboard.

That's why today we’re adding more than 100 virtual keyboards, transliteration and IMEs—collectively called input tools—in Gmail. These tools enable you to type in the language and keyboard layout you’re accustomed to, making it easy to keep in touch with family, friends and coworkers from any computer. You can even switch between languages with one click.


To try it out, check the box next to Enable input tools under Language in Settings.

Once you’ve enabled it, you’ll see the Input Tools icon next to the Settings button in your toolbar, and you can turn on and off any Input Tool from there.


With these new virtual keyboards, Gmail supports typing in 75 languages—a big jump from the five languages that were initially supported when we introduced Indic transliteration in Gmail in 2009.

Gmail’s users are from all over the world—and language should never get in the way of a good conversation. If you'd like to use Input Tools in other places, try out the Chrome extension, the Windows desktop client or the Android app.



(Cross-posted on the Gmail and Enterprise Blogs)

Friday, October 5, 2012

ReFr: A New Open-Source Framework for Building Reranking Models



We are pleased to announce the release of an open source, general-purpose framework designed for reranking problems, ReFr (Reranker Framework), now available at: http://code.google.com/p/refr/.

Many types of systems capable of processing speech and human language text produce multiple hypothesized outputs for a given input, each with a score. In the case of machine translation systems, these hypotheses correspond to possible translations from some sentence in a source language to a target language. In the case of speech recognition, the hypotheses are possible word sequences of what was said derived from the input audio. The goal of such systems is usually to produce a single output for a given input, and so they almost always just pick the highest-scoring hypothesis.

A reranker is a system that uses a trained model to rerank these scored hypotheses, possibly inducing a different ranked order. The goal is that by employing a second model after the fact, one can make use of additional information not available to the original model, and produce better overall results. This approach has been shown to be useful for a wide variety of speech and natural language processing problems, and was the subject of one of the groups at the 2011 summer workshop at Johns Hopkins’ Center for Language and Speech Processing. At that workshop, led by Professor Brian Roark of Oregon Health & Science University, we began building a general-purpose framework for training and using reranking models. The result of all this work is ReFr.

From the outset, we designed ReFr with both speed and flexibility in mind. The core implementation is entirely in C++, with a flexible architecture allowing rich experimentation with both features and learning methods. The framework also employs a powerful runtime configuration mechanism to make experimentation even easier. Finally, ReFr leverages the parallel processing power of Hadoop to train and use large-scale reranking models in a distributed computing environment.

Wednesday, October 3, 2012

EMEA Faculty Summit 2012



Last week we held our fifth Europe, Middle East and Africa (EMEA) Faculty Summit in London, bringing together 94 of EMEA’s foremost computer science academics from 65 universities representing 25 countries, together with more than 60 Googlers.

This year’s jam-packed agenda included a welcome reception at the Science Museum (plus a tour of the special exhibition: “Codebreaker - Alan Turing’s life and legacy”), a keynote on “Research at Google” by Alfred Spector, Vice President of Research and Special Initiatives and a welcome address by Nelson Mattos, Vice President of Engineering and Products in EMEA, covering Google’s engineering activity and recent innovations in the region.

The Faculty Summit is a chance for us to meet with academics in Computer Science and other areas to discuss the latest exciting developments in research and education, and to explore ways in which we can collaborate via our our University Relations programs.

The two and a half day program consisted of tech talks, break out sessions, a panel on online education, and demos. The program covered a variety of computer science topics including Infrastructure, Cloud Computing Applications, Information Retrieval, Machine Translation, Audio/Video, Machine Learning, User Interface, e-Commerce, Digital Humanities, Social Media, and Privacy. For example, Ed H. Chi summarized how researchers use data analysis to understand the ways users share content with their audiences using the Circle feature in Google+. Jens Riegelsberger summarized how UI design and user experience research is essential to creating a seamless experience on Google Maps. John Wilkes discussed some of the research challenges - and opportunities - associated with building, managing, and using computer systems at massive scale. Breakout sessions ranged from technical follow-ups on the talk topics to discussing ways to increase the presence of women in computer science.

We also held one-on-one sessions where academics and Googlers could meet privately and discuss topics of personal interest, such as how to develop a compelling research award proposal, how to apply for a sabbatical at Google or how to gain Google support for a conference in a particular research area.

The Summit provides a great opportunity to build and strengthen research and academic collaborations. Our hope is to drive research and education forward by fostering mutually beneficial relationships with our academic colleagues and their universities.

Tuesday, October 2, 2012

Celebrating teachers who make a difference with Google

For most of us, there’s at least one teacher whose name we will never forget—that favorite teacher who made a difference in our education, whether they were our first grade art teacher or a professor in college. For me, that teacher was Ms. Taylor, my 8th grade science teacher. Ms. Taylor didn't just foster my love of science—she understood that 8th grade can be a tough time for students as they try to navigate social cliques and prepare for the pressure of high school. Ms. Taylor knew that taking the time to ask us if we were feeling okay was just as important as teaching us about geological formations. She didn’t just care about teaching us—she genuinely cared about us as people.

This Friday is World Teachers' Day, and we want to honor the teachers like Ms. Taylor who helped make us the people we are today. We’ve long supported education through technology, offering free tools like YouTube Edu and Google Apps for Education, and by developing cost-efficient devices like Chromebooks. But it’s the teachers who really make the difference by creatively incorporating that technology into their classrooms. As technology usage in schools increases, we hear even more amazing stories about how teachers and students are using our products to foster collaborative learning.

And that usage is growing quickly. As of today, more than 20 million students, faculty and staff worldwide use Google Apps for Education. In addition, in the last year we announced that:
  • 400+ universities are posting lectures and/or full courses online using YouTube Edu
  • 600,000 staff from the Philippines Department of Education will now be using Google Apps
  • Universities across the continents are signing up for Apps, including schools in Poland, Spain, the Netherlands and Africa
  • More than 500 schools and districts went back to school with Chromebooks this fall
  • Seven of the eight Ivy League universities and 72 of this year’s top 100 U.S. universities (as determined by 2013 U.S. News and World Report’s ranking) have gone Google with Google Apps for Education
As a tribute to the educators who are putting these tools to work, this week we’ll be highlighting a few amazing teachers on our Google in Education page on Google+. To kick off the series, we want to celebrate Ms. Kornowski—a science teacher at Kettle-Moraine High School in Wales, WI, who is using Google Forms to bring her students together.



To all the Ms. Taylors and Ms. Kornowskis out there—thank you, both for the positive impact you have on your students and for letting Google be a part of that experience.