Google Data » Andrew

A new look for Google Translate for Android

Andrew — Wed, 12 Jan 2011 19:08:00 +0000

When we launched the first version of Google Translate for Android in January 2010, we were excited about the year ahead. For the first time, we were bringing the capabilities supported on Google Translate—like machine translation, romanization of non-Roman scripts and spoken translations—to the Android platform. We also offered voice input to let you speak the word or phrase you wanted to translate instead of typing it in, and SMS translation so you could translate SMS messages sent to you in foreign languages.

Today, we’re refreshing Translate for Android with several updates to make the app easier to interact with. Among other improvements, we’ve created better dropdown boxes to help select the languages you want to translate from and into, an improved input box, and cleaner icons and layout.

We also want to let you in on an experimental feature that’s still in its earliest stages—Conversation Mode. This is a new interface within Google Translate that’s optimized to allow you to communicate fluidly with a nearby person in another language. You may have seen an early demo a few months ago, and today you can try it yourself on your Android device.

Currently, you can only use Conversation Mode when translating between English and Spanish. In conversation mode, simply press the microphone for your language and start speaking. Google Translate will translate your speech and read the translation out loud. Your conversation partner can then respond in their language, and you’ll hear the translation spoken back to you. Because this technology is still in alpha, factors like regional accents, background noise or rapid speech may make it difficult to understand what you’re saying. Even with these caveats, we’re excited about the future promise of this technology to be able to help people connect across languages.

As Android devices have spread across the globe, we’ve seen Translate for Android used all over. The majority of our usage now comes from outside the United States, and we’ve seen daily usage from more than 150 countries, from Malaysia to Mexico to Mozambique. It’s really rewarding for us to see how this new platform is helping us break down language barriers the world over.

Translate supports 53 languages, from Afrikaans to Yiddish, and voice input for 15 languages. You can download the application, available for devices running Android 2.1 and above, by searching for “Google Translate” in Android Market or by scanning the QR Code below.

Posted by Awaneesh Verma, Product Manager

Going Global with Google

Andrew — Tue, 07 Dec 2010 01:11:00 +0000

(Cross-posted from the Inside AdWords blog)

AdWords started with the idea of connecting you with customers searching for your goods and services - wherever those customers may be. Today, we're announcing the launch of Google Global Market Finder, a free online tool that helps you find new markets overseas, and Google Ads for Global Advertisers, a new website that brings together Global Market Finder, AdWords, Google Translate, and other tools that can help you find, engage with, and support your customers worldwide - whether those customers are consumers or businesses searching for suppliers online.

Global Market Finder helps you answer the question: who are my potential customers overseas?

click for full size image

To use the tool, enter keywords that describe your product or service and select a market or region you'd like to explore. You can choose from regions such as the European Union, the “G20” economies, or the Americas.

Global Market Finder automatically translates your keyword into languages used in each of your selected markets. It then ranks each location by market opportunity by combining search volume, suggested bid price, and competition for each translated keyword. With this tool, businesses can answer questions like “how competitive is this market?”, “how does demand in one location compare to demand elsewhere in the world?”, and “how much would it cost to start advertising in this new market?” Since automatic translation is not perfect, be sure to confirm that the translated terms are appropriate for your business.

The second resource we're launching today is Google Ads for Global Advertisers, a website where you can learn more about Google tools that can help expand your business overseas. Google Ads for Global Advertisers contains step-by-step guides and tools that take you from local to global: from finding the right markets to expand your business, to localizing your website and campaigns into another language, to running ads in a foreign market, and finally, to monitoring your global ad spend.

click for full size image

This website pulls together resources for you to:

Find the right market for your products and services, by using tools such as the Global Market Finder
Translate your websites and ad text using Google Translate Web Element and Google Translator Toolkit
Reach new customers with relevant online ads
Understand options for international payment, shipping, and customer service

On the website, you will also find examples of businesses like yours that have gone global using AdWords - examples such as a mosaic company in Lebanon, a bespoke shoe retailer in Sydney, a tech support company in India, and a bed and breakfast in Poland. We hope our new website and tools will help you reach overseas markets, whether you’re a small business testing exports for the first time, or a mid-sized company looking to grow your multinational business.

Both Global Market Finder and Google Ads for Global Advertisers will be available in 43 languages. There are more than 1.9 billion people online. Wouldn’t you like to add some of them as customers?

Posted by Michael Galvez and Srinidhi Viswanatha, Global Advertisers team

Welcome, Google Apps users!

Andrew — Thu, 02 Dec 2010 20:24:00 +0000

Google Apps recently launched an improvement that made dozens of exciting Google services available to Google Apps users for the first time. As part of this launch, Google Translator Toolkit is now available to our Google Apps users for free with their Apps accounts.

Google Apps is Google’s suite of cloud-based messaging and collaboration apps used by over 30 million users in small businesses, large enterprises, educational institutions, government agencies and non-profit organizations around the world. If your organization hasn’t gone Google yet you can learn more about how to lower IT costs and improve productivity and collaboration with Google Apps.

For those users who have a Google Apps account, if your administrator has already transitioned your organization to the new infrastructure, you can now use Translator Toolkit by signing in at translate.google.com/toolkit with your existing Apps account.

For more details, read the complete post on the Google Enterprise blog and follow all the updates on other newly-available services for Google Apps users.

Posted by Jeremiah Dillon, Product Marketing Manager

There are some things Google can’t translate (yet)

Andrew — Tue, 16 Nov 2010 23:46:00 +0000

Google Translate may be able to translate over 50 languages, everything from Afrikaans to Yiddish, but there are some things we still can’t translate. A baby babbling, for example:

For the week of November 15th we are releasing five videos of things Google can’t translate (at least not yet)! Check out the videos and share them with your friends. If you can think of other things you wish Google translated (like your calculus homework or your pet hamster), tweet them with the tag #GoogleTranslate. We’ll be making a video of at least one of the suggestions and adding it to our page.

So join us as we laugh about the things Google can’t translate (yet), and celebrate the over 50 languages it can!

Posted by Andrew Gomez, Product Marketing Manager

There are some things Google can’t translate (yet)

Andrew — Tue, 16 Nov 2010 23:46:00 +0000

Endangered Languages to Endure on YouTube

Andrew — Thu, 28 Oct 2010 16:30:00 +0000

(Cross-posted from the Google.org blog)

Many of the world's smallest and most endangered languages have no written form and have never been recorded or scientifically documented. Today, the National Geographic Enduring Voices YouTube channel will launch and allow many of these tongues to have a presence on the Internet for the very first time. Linguists Dr. K. David Harrison and Dr. Gregory Anderson from the Living Tongues Institute have teamed up with Google.org to allow small and endangered languages that may have never been heard outside of a remote village to reach a global audience. Using YouTube as a platform, researchers, academics and communities can now collaborate more effectively on promoting language revitalization.

The YouTube channel features videos such as hip-hop performed by Songe Nimasow in the Aka language of India, songs by Aydyng Byrtan-ool, a talented young Tuvan singer and epic storyteller in Southern Siberia, and videos demonstrating how the Foe language of Papua New Guinea uses body parts to count from 1 to 37.

The launch of the channel comes on the heels of an announcement by Harrison and Anderson of a “hidden” language of India, known locally as Koro, that is new to science and had never been documented outside of its rural community. Koro is one of half of the world’s languages likely to vanish in the next 100 years.

In addition to using YouTube to help revitalize endangered and minority languages, communities can also take advantage of Google Translator Toolkit that announced the addition of 284 new languages last year to make translation faster and easier.

In the midst of a language extinction crisis, we are also seeing a global grassroots movement for language revitalization. Speakers are leveraging new technologies, such as social networking and YouTube, to sustain small languages. As Harrison describes in his book "The Last Speakers," we are all impoverished when a language dies, and all enriched by the human knowledge base found in the world's smallest tongues.

Learn more about Harrison and Anderson's efforts to document languages through the Enduring Voices Project.

Posted by Kirsten Olsen Cahill, Product Marketing Manager, Google.org

Endangered Languages to Endure on YouTube

Andrew — Thu, 28 Oct 2010 16:30:00 +0000

Poetic Machine Translation

Andrew — Tue, 05 Oct 2010 22:33:00 +0000

Once upon a midnight dreary, long we pondered weak and weary,

Over many a quaint and curious volume of translation lore.

When our system does translation, lifeless prose is its creation;

Making verse with inspiration no machine has done before.

So we want to boldly go where no machine has gone before.

Quoth now Google, "Nevermore!"

Robert Frost once said, “Poetry is what gets lost in translation”. Translating poetry is a very hard task even for humans, and is clearly beyond the capability of current machine translation systems. We therefore, out of academic curiosity, set about testing the limits of translating poetry and were pleasantly surprised with the results!

We will present a paper on poetry translation at the EMNLP conference this year. In this paper, we investigate the purely technical challenges around generating translations with fixed rhyme and meter schemes.

The value of preserving meter and rhyme in poetic translation has been highly debated. Vladimir Nabokov famously claimed that, since it is impossible to preserve both the meaning and the form of the poem in translation, one must abandon the form altogether. Another authority (and for us, computer scientists, perhaps the more familiar one), Douglas Hofstadter argues that preserving the form is very important to maintaining the feeling and the sound of a poem. It is in this spirit that we decided to experiment with translating not only poetic meaning, but form as well.

A Statistical Machine Translation system, like Google Translate, typically performs translations by searching through a multitude of possible translations, guided by a statistical model of accuracy. However, to translate poetry, we not only considered translation accuracy, but meter and rhyming schemes as well. In our paper we describe in more detail how we altered our translation model, but in general we chose to sacrifice a little of the translation’s accuracy to get the poetic form right.

As a pleasant side-effect, the system is also able to translate anything into poetry, allowing us to specify the genre (say, limericks or haikus), or letting the system pick the one it thinks fits best. At the moment, the system is too slow to be made publicly accessible, but we thought we’d share some excerpts:

A stanza from Essai monographique sur les Dianthus des Pyrénées françaises by Edouard Timbal-Lagrave and Eugène Bucquoy, translated to English as a pair of couplets in iambic tetrameter:

So here's the dear child under land,

will not reflect her beauty and

besides the Great, no alter dark,

the pure ray, fronts elected mark.

Voltaire’s La Henriade, translated as a couplet in dactylic tetrameter:

These words compassion forced the small to lift her head

gently and tell him to whisper: “I'm not dead."

Le Miroir des simples âmes, an Old French poem by Marguerite Porete, translated to Modern French by M. de Corberon, and then to haiku by us:

“Well, gentle soul”, said

Love, “say whatever you please,

for I want to hear.”

More examples and technical details can be found in our research paper (as well as clever commentary).

Posted by Dmitriy Genzel, Software Engineer

Poetic Machine Translation

Andrew — Tue, 05 Oct 2010 22:33:00 +0000

Once upon a midnight dreary, long we pondered weak and weary,

Over many a quaint and curious volume of translation lore.

When our system does translation, lifeless prose is its creation;

Making verse with inspiration no machine has done before.

So we want to boldly go where no machine has gone before.

Quoth now Google, "Nevermore!"

A stanza from Essai monographique sur les Dianthus des Pyrénées françaises by Edouard Timbal-Lagrave and Eugène Bucquoy, translated to English as a pair of couplets in iambic tetrameter:

So here's the dear child under land,

will not reflect her beauty and

besides the Great, no alter dark,

the pure ray, fronts elected mark.

Voltaire’s La Henriade, translated as a couplet in dactylic tetrameter:

These words compassion forced the small to lift her head

gently and tell him to whisper: “I'm not dead."

Le Miroir des simples âmes, an Old French poem by Marguerite Porete, translated to Modern French by M. de Corberon, and then to haiku by us:

“Well, gentle soul”, said

Love, “say whatever you please,

for I want to hear.”

More examples and technical details can be found in our research paper (as well as clever commentary).

Posted by Dmitriy Genzel, Software Engineer

Veni, Vidi, Verba Verti

Andrew — Fri, 01 Oct 2010 17:43:00 +0000

[We’ve added Latin as an alpha language to translate.google.com! Alpha languages aren’t perfect, but we think the addition will help unlock many classic Latin texts and documents. Learn more from our programmer Jakob in the post below. Don’t speak Latin? Good thing there is now an easy way to translate the language...]

Ut munimenta linguarum convellamus et scientiam mundi patentem utilemque faciamus, Ut munimenta linguarum convellamus et scientiam mundi patentem utilemque faciamus, instrumenta convertendi multarum nationum linguas creavimus. Hodie nuntiamus primum instrumentum convertendi linguam qua nulli nativi nunc utuntur: Latinam. Cum pauci cotidie Latine loquantur, quotannis amplius centum milia discipuli Americani Domesticam Latinam Probationem suscipiunt. Praeterea plures ex omnibus mundi populis Latinae student.

Hoc instrumentum convertendi Latinam rare usurum ut convertat nuntios electronicos vel epigrammata effigierum YouTubis intellegamus. Multi autem vetusti libri de philosophia, de physicis, et de mathematica lingua Latina scripti sunt. Libri enim vero multi milia in Libris Googlis sunt qui praeclaros locos Latinos habent.

Convertere instrumentis computatoriis ex Latina difficile est et intellegamus grammatica nostra non sine culpa esse. Autem Latina singularis est quia plurimi libri lingua Latina iampridem scripti erant et pauci novi posthac erunt. Multi in alias linguas conversi sunt et his conversis utamur ut nostra instrumenta convertendi edoceamus. Cum hoc instrumentum facile convertat libros similes his ex quibus edidicit, nostra virtus convertendi libros celebratos (ut Commentarios de Bello Gallico Caesaris) iam bona est.

Proximo tempore locum Latinum invenies vel auxilio tibi opus eris cum litteris Latinis, conare hunc.

Jakob Uszkoreit, Ingeniarius Programmandi et Ben Bayer, Magister Spatii et Temporis

Share your story with the new Google Translate

Andrew — Mon, 30 Aug 2010 22:05:00 +0000

Today, you may have noticed a brighter looking Google Translate. We’re currently rolling out several changes globally to our look and feel that should make translating text, webpages and documents on Google Translate even easier. These changes will be available globally within a couple of days.

Google Translate’s shiny new coat of paint

With today’s functional and visual changes we wanted to make it simpler for you to discover and make the most of Google Translate’s many features and integrations. For example, did you know that you can search across languages on Google using Google Translate? Or that you can translate incoming email in Gmail or take Google Translate with you on your phone? We’ve added all these tips on the new Do more with Google Translate page. You can also see some of these tips rotating on the new homepage.

We’ve also created an Inside Google Translate page, where you can learn how we create our translations. Is it the work of magic elves or learned linguists? Here Anton Andryeyev, an engineer on our team, gives you the inside scoop:

It’s always inspiring for us to learn how Google Translate enables people to break down communication barriers around the world. Lisa J. recently shared with us how she uses Google Translate to stay in touch with her grandparents. “I moved to the U.S. from China when I was six,” Lisa told us, “so I speak both English and Chinese fluently but I’m not very good at reading the complex Chinese alphabet.” When she gets an email from her grandparents in China, Google Translate helps her understand the sentences she can’t quite read. She also uses Google Translate when she’s writing her response. “I use Google Translate to make sure I’m using the right character in the right place,” she explained.

Do you use Google Translate to stay in touch with distant relatives? Read foreign news? Or make the most of your vacation? We’d love to hear from you, and invite you to share your story with us. Who knows, we might feature your story on the Google Translate blog!

UPDATED 9/3: Thank you for your helpful feedback in the comments. Please add your feedback in our Google Translate Help Forum for us to easily track and respond to. Thank you!

Posted by Awaneesh Verma, Product Manager

Translate faster with keyboard shortcuts in Translator Toolkit

Andrew — Wed, 04 Aug 2010 20:40:00 +0000

Speed is an important part of translation, whether you’re a professional rushing to meet a client deadline or a volunteer looking to spend less time on a mouse and more on your keyboard. Last month, we added a dozen new shortcuts to Translator Toolkit, so you can spend less time taking your hands off your keyboard to use the mouse.

Translator Toolkit’s 18 keyboard shortcuts help you quickly navigate through your translation or call up standard edit functions. For example, you can use Ctrl+F to start Find and Replace or Ctrl+Home to jump to the beginning of the document.

Use Ctrl+F to start Find and Replace

We’ve also added shortcuts for advanced users, including automatically replacing segments with source (Ctrl+Shift+S), machine translation (Ctrl+Shift+M), or translation memory matches (Ctrl+Shift+L). You can also use Ctrl+Shift+C to start concordance search or Ctrl+Shift+I to automatically insert placeholders.

Use Ctrl+Shift+C to start concordance

Check out these improvements now in Translator Toolkit. We'd love to hear what you think.

Posted by Sunil Chandra, Engineer

Translate faster with keyboard shortcuts in Translator Toolkit

Andrew — Wed, 04 Aug 2010 20:40:00 +0000

Use Ctrl+F to start Find and Replace

Use Ctrl+Shift+C to start concordance

Check out these improvements now in Translator Toolkit. We'd love to hear what you think.

Posted by Sunil Chandra, Engineer

Translating Wikipedia

Andrew — Wed, 14 Jul 2010 22:22:00 +0000

We believe that translation is key to our mission of making information useful to everyone. For example, Wikipedia is a phenomenal source of knowledge, especially for speakers of common languages such as English, German and French where there are hundreds of thousands—or millions—of articles available. For many smaller languages, however, Wikipedia doesn’t yet have anywhere near the same amount of content available.

To help Wikipedia become more helpful to speakers of smaller languages, we’re working with volunteers, translators and Wikipedians across India, the Middle East and Africa to translate more than 16 million words for Wikipedia into Arabic, Gujarati, Hindi, Kannada, Swahili, Tamil and Telugu. We began these efforts in 2008, starting with translating Wikipedia articles into Hindi, a language spoken by tens of millions of Internet users. At that time the Hindi Wikipedia had only 3.4 million words across 21,000 articles––while in contrast, the English Wikipedia had 1.3 billion words across 2.5 million articles.

We selected the Wikipedia articles using a couple of different sets of criteria. First, we used Google search data to determine the most popular English Wikipedia articles read in India. Using Google Trends, we found the articles that were consistently read over time––and not just temporarily popular. Finally we used Translator Toolkit to translate articles that either did not exist or were placeholder articles or “stubs” in Hindi Wikipedia. In three months, we used a combination of human and machine translation tools to translate 600,000 words from more than 100 articles in English Wikipedia, growing Hindi Wikipedia by almost 20 percent. We’ve since repeated this process for other languages, to bring our total number of words translated to 16 million.

We’re off to a good start but, as you can see in the graph below, we have a lot more work to do to bring the information in Wikipedia to people worldwide:

Number of non-stub Wikipedia articles by Internet users, normalized (English = 1)

We’ve also found that there are many Internet users who have used our tools to translate more than 100 million words of Wikipedia content into various languages worldwide. If you do speak another language we hope you’ll join us in bringing Wikipedia content to other languages and cultures with Translator Toolkit.

We presented these results last Saturday, July 10, at Wikimania 2010 in Gdańsk, Poland. We look forward to continuing to support the creation of the world’s largest encyclopedia and we can’t wait to work with Wikipedians and volunteers to create more content worldwide.

Posted by Michael Galvez, Product Manager

The Polish Ministry of Economy goes multi-lingual with Google Translate

Andrew — Wed, 23 Jun 2010 21:44:00 +0000

The Google Translate team is always thinking about how to make information accessible to all internet users, not matter what language it is written in. Often people come across webpages in foreign languages, but don’t know that they can go to Google Translate to find out what the page says. This problem is especially challenging for government institutions that have wide audiences and information that needs to be translated into multiple languages. The Polish Ministry of Economy recently solved this problem by using our website translation element to make their webpage multilingual for visitors who don’t speak Polish.

By embedding the website translation element, the Ministry enabled visitors to translate site text instantly without leaving the Ministry’s webpage. With the addition of the Google Translate drop-down box, their webpages are now accessible to Chinese, English, French, German, Russian, and Spanish speakers who don’t have any knowledge of Polish at all!

While the translation quality is not perfect, it is good enough to allow visitors who don’t speak Polish to interact in a way they never could have before. Check out the Ministry’s site or add the website translation element to your own webpage!

Posted by Andrew Gomez, Associate Product Marketing Manager

The Polish Ministry of Economy goes multi-lingual with Google Translate

Andrew — Wed, 23 Jun 2010 21:44:00 +0000

Create your own Search Story!

Andrew — Tue, 13 Apr 2010 18:47:00 +0000

If you live in the United States you probably saw our Parisian Love ad during the Super Bowl that famously highlighted Google Translate in the search results ("translate tu es très mignon"). Many users responded by making their own parody videos covering topics they felt passionate about.

We’re excited to say that today we launched a tool that allows any user to quickly and easily make their own Search Story video. You just have to put in the search queries you want and the tool does the rest! Check out this video that highlights translate in Google search:

As you can see, by typing “translate _______” into Google search you can see translations right on the search results page--as well as tell a pretty funny story. What sort of videos will you make?

Posted by Andrew Gomez, Associate Product Marketing Manager

More love for our foreign language learners

Andrew — Mon, 12 Apr 2010 20:41:00 +0000

With Google Translate we are always iterating, adding new features and improving translation quality. Today we would like to highlight a few of the features we’ve added recently.

In November 2009 we launched a “text-to-speech” feature which allowed users to hear words and sentences in English read out loud by Google Translate. Since then we’ve added Haitian Creole and in March added French, Italian, and German to the list of supported languages. To hear any of languages out loud, just click the icon next to the translation when you visit translate.google.com.

In addition, we have recently doubled the number of languages that support transliterated input. Last year we announced that you could type transliterated Arabic, Hindi, and Persian. We have now extended that functionality to Greek, Russian and Serbian. With all of these languages you can type transliterated words in roman script and have the script converted automatically. Go ahead and check it out!

While we are on the topic of transliteration, the labs product Google Transliteration has recently added Hebrew, Oriya and Sinhalese to its supported languages. We expect to have Hebrew transliteration working in Google Translate soon as well.

Added on 4/14/10: Hindi and Spanish "text-to-speech" are now live on Google Translate as well.

Posted by Jeff Chin, Product Manager

An experiment in cross-language communication with the BBC

Andrew — Thu, 25 Mar 2010 18:12:00 +0000

Earlier this month BBC launched an experiment using Google Translate to faciliate real-time discussion across languages. The project was part of the BBC’s SuperPower series which explored the transformative power of the internet. On SuperPower Nation Day, BBC readers from around the world were invited to discuss the Nation Day event online--and have their comments translated live for others to read.

The initiative used Google Translate to translate comments between seven languages: Arabic, Chinese, English, Persian, Indonesian, Portugese, and Spanish. While the translations were not perfect, most users were surprised by how well Google Translate enabled people from around the world to communicate with one another.

“Acredito que isto pode dar certo!” wrote Nathana from Brazil in one of the comments--or, as translated by Google Translate, “I believe this can work!”

While a few of the translations looked strange at first glance, most users found them intelligible and useful. For example one user’s comment was translated from Persian into, “World, the village is small. We all need each other to help maintain the world.” As you can see, not perfect, but understandable.

This experiment is a good example of how Google Translate can be used to help people of different languages communicate with one another. While there is still a long way to go in improving Google’s translation technology, we hope and expect to see projects like the BBC’s SuperPower Nation Day become more and more common--and as the opportunities arise we look forward to making our contribution with Google Translate.

Posted by Jeff Chin, Product Manager

Translating Youtube with auto-captioning

Andrew — Tue, 23 Mar 2010 00:13:00 +0000

We recently mentioned on the official Youtube blog that we were enabling auto-captioning for all Youtube videos. This new technology takes advantage of Google’s speech-to-text algorithms to add captions to videos with spoken English content. Captions have clear benefits for the hearing impaired, but there is another benefit which is worth noting: translation.

As you know, Google Translate can already translate between any of our 52 supported languages, whether the text is a word, paragraph or website. With this new advancement in Youtube’s technology we can now add “captions” to that list as well.

Currently Youtube’s auto-captioning technology can only transcribe spoken English but, thanks to Google Translate, these English captions can be translated for non-English speakers as well. Check out the caption and translation technology in this video of Barack Obama speaking to students or author Elizabeth Gilbert speaking at TED. Hover over the “caption” icon and select “Translate Captions” to activate the translations.

Both the auto-captioning and translation technologies are generated by machines, so there are still plenty of situations where mistakes will occur. Furthermore, when these technologies work together, mistakes can be amplified so you may find some captions that don’t make much sense. We know there is a lot of room to improve both technologies, but we feel that offering this translation feature is an important step in making information on Youtube more accessible.

In the next few months we expect over 150,000 Youtube channels to implement auto-captioning with translation. This is just the beginning and we hope that all Youtube content will soon be enjoyed by all Youtube users, regardless of what language they speak.

Posted by Andrew Gomez, Associate Product Marketing Manager

A faster, simpler, and safer browser goes polyglot

Andrew — Wed, 17 Mar 2010 17:30:00 +0000

If you're excited about the latest in integrating translation tools into your web experience, today's new stable release of Google Chrome brings to our users translation capabilities in the web browser. Here's quick demo on how the translation feature works in Chrome:

When the language of the webpage you're viewing is different from your preferred language setting, Chrome will display an infobar asking if you'd like the page to be translated for you, using Google Translate. With just a click, the entire text on the page will be translated into your language of preference, without the need for browser extensions or plugins. If you don't want Chrome to offer to translate a particular language or web page, you can control these settings by clicking on the "Options" button in the infobar.

So how does the browser actually 'know' what language the page is in? Language detection takes place locally on your computer, so no information is sent to Google Translate until you choose to translate a page. Language detection in Chrome is based on the compact language detection library (CLD), which we've made available as open source code.

For those of you who are interested in the technical nitty-gritty, here's what takes place under the hood: for most languages, the CLD determines the language of a page by breaking down its text in quadgrams, or sequences of up to four characters. The CLD then looks up each quadgram in a large hashtable that contains language probabilities, which is included in the Chrome binaries. This hashtable was originally built by processing language probabilities over billions of web pages that are indexed by Google's search engine. In just a few milliseconds, the CLD can accurately determine the language of most web pages. Chrome shows an infobar offering to translate the page only when the CLD has detected the language of a web page above a certain degree of confidence. If you click the "Translate" button in the infobar, the text contained in the page is then sent to Google Translate's servers (over a secure connection if the page was served over HTTPS). Thanks to the work of the Google Translate team, Google Translate's servers return this translated text quickly so that Chrome can replace the text in the page with the translated version. Rest assured, the request to Google Translate's servers does not include any cookies.

We're excited to introduce translation in the browser, and look forward to improving this feature over future releases of Chrome. For example, we hope to work on include making language detection even more precise -- by providing larger CLD tables without increasing the size of the browser's installation package, and improving the way Chrome interacts with other website translation tools.

With this new stable release of Chrome, you can easily read a diversity of foreign language information sources, access educational materials from universities around the world, and even conduct online commerce across borders and languages -- all in your native language. We hope that the browser can truly be a passport to a web that is remarkably local as it is global. You can try translation in the browser for yourself by downloading Google Chrome at google.com/chrome. For those of you who aren't yet acquainted with Chrome, you can learn more about Chrome's speed, security, and many other features.

Posted by Jay Civelli, Software Engineer, Google Chrome

Welcome to the Google Translate Blog

Andrew — Tue, 16 Feb 2010 18:35:00 +0000

Welcome to our new blog for Google Translate. You're probably familiar with Google Translate, our free service that offers automatic machine translations between 52 languages — more than 2500 language pairs. In the last two years we've made many improvements to Translate, including the most recent additions of Haitian Creole and Persian , transliteration for reading and writing, and integration with Google Toolbar to read any website — just to name a few. There's always a lot happening related to Translate, so we've created this blog to share all of our new developments with the world.

When I joined the Google research team in 2004, we offered machine translation for fewer than 10 languages. By significantly increasing translation quality and coverage, we believed that Google Translate could be a valuable tool for breaking down language barriers that limit communication and access to information for many people. We therefore set out to build a new machine translation system leveraging Google's strengths: large data resources and the corresponding computational power to process it. To achieve this goal, we built advanced machine learning algorithms to translate content automatically based on an extensive corpus of translated texts.

In 2006 we launched Chinese and Arabic, the first languages based on this new technology. Since then we've continually improved quality and added more languages to reach our current total of 52 different languages — representing well over 95 percent of the Internet users. We're especially proud of our ability to serve very small language communities by offering Icelandic, Yiddish, Albanian, Catalan, Welsh and Haitian Creole translation.

Thanks to these advancements millions of people now use Google Translate and usage continues to grow significantly. People around the world can search the web in most languages and read websites, blogs and tweets regardless of the language they're written in. In short, we're beginning to see traditional language barriers break down. We believe that translation has the power not only to bring people together across cultures but also to facilitate progress and positive change.

Our mission is to make all information accessible to you, no matter what language you speak or what device or platform you use. Imagine having a phone conversation with a person half-way across the globe and being able to understand them — no matter what language they're speaking — as the conversation is translated in real-time. Our universal translator hasn't reached all the capabilities you know from Star Trek quite yet but we're definitely headed in that direction. Universally accessible and accurate translation is a lofty goal, but we're committed to it.

So once again, welcome to our blog. We hope you'll come back often for all the latest news on Google Translate.

Posted by Franz Och, Principal Scientist

1 Billion Messages Not Served

Andrew — Thu, 13 Dec 2007 18:16:00 +0000

Posted by Adam Swidler, Product Marketing Manager, Postini

Postini is a recent addition to Google that offers solutions that help enterprises make their existing email infrastructure more secure, compliant and productive. We process email for more than 35,000 businesses and 12 million end users, and block about 1 billion messages per day, which is a good sample size to report on global spam trends for businesses. In 2007, Postini data centers recorded the highest levels of spam and virus attacks in history. Much of this was fueled by an increase in the number of botnet computers being used to send spam. Botnets are networks of infected PCs, usually with broadband Internet connections that are co-opted by hackers and used to send spam and virus attacks. Often they are compromised without their owner's knowledge. We started to see these botnets kick in back in September of 2006. Since that time, spam volumes are up more than 163 percent. We saw a peak of activity in October 2007 where volume was a 263 percent increase from September 2006 and Postini blocked 47 billion spam messages, more than 320 Terabytes of spam (now that's a lot of spam). The average unprotected e-mail user would have received 32,000 spam messages in their in-boxes so far this year. Talk about lost productivity. In fact, Nucleus research estimates unchecked spam can cost a company up to $742 per user.

But what's really different this year is the innovation with which spammers attempted to evade detection by spam filters. In the early part of 2007, image spam was used heavily, with the spam content (such as "pharmaceuticals for sale," "hot stocks," etc) contained in an image attached to the message. Over the course of the year image spam declined and was replaced by PDF spam, document and spreadsheet spam and even multimedia spam. That's right - an audio file promoting a particular stock. We saw examples of compressed and password protected emails as well. All this effort to deliver spam content in email attachments had a significant impact on the size of spam overall. Taking 7.5 Kb as an average spam message size, an organization with 100 employees (that didn't use a hosted solution to block spam outside the firewall) would have wasted 22Gb of storage and bandwidth. Who wants that sitting on their servers?

The chart below shows the trend of the volume of spam rising throughout the year (blue line) and the peaks in the size of spam (orange line):

In Europe, spam volume was also up in 2007 and the percentage of email that is spam increased from 70% to 90% as shown below:

Virus attacks in 2007 were also at record levels and showed similar techniques to what we observed for spam email. In January 2007, the high profile "Storm" botnet got its start with an email that was spammed out with an executable file attached. In April and May, we saw virus emails with password protected executables attached and the password contained in the body of the email. In the late summer, we saw a huge spike of Storm virus attacks that used a blended threat - an email message with a URL that took the user to an infected website that then downloaded the Storm malware to the PC. The linkage between spam and viruses continued in 2007, with messages being spammed out from botnets with virus attacks intended to add more computers to the botnet. The chart below shows the Storm virus activity for 2007:

For 2008, this game of chess will continue, and the stakes may become even higher. While the number of threats may not increase, the complexity will. Businesses will be challenged to identify more types of malicious content and protect sensitive information against new methods of social engineering. So what does the forecast for next year looks like? Here are some of the trends we expect to see coming to a business near you in 2008:

Spam volume will stabilize and could actually decrease in 2008 as spam attacks become more targeted and less of a pure volume game. As more and more spam content will be contained in attachments, we do expect that the overall size of spam will continue to increase.

Virus attacks will continue to blend with spam and will focus more on identity theft. They will utilize increasingly sophisticated social engineering techniques that will be related to specific current events such as the Olympics, the Super Bowl, natural disasters etc. Virus attacks will become more targeted toward executives at specific companies and will appear to come from legitimate business agencies. Their goal will increasingly be to steal corporate and government data. We expect to see several of these types of attacks, leading to data breaches from commercial enterprises and government agencies. This may force some companies to modify their email practices, such as financial institutions not including any links in their email communications to customers.

More businesses and organizations will implement specific policies that address outbound content in email and will deploy systems to monitor and enforce those policies to prevent sensitive or confidential data leaks.

The growing need for managing consumer data privacy and retention policies globally will drive growth of encryption and archiving and hosted solutions will play a major role in reducing the cost and complexity.

Identity theft attacks will increasingly be launched through web sites, especially those that enable user created content such as social networking sites, blogs and auction sites.

If you’d like to know more about Postini’s spam and virus trends, you can read our weekly summaries on the Postini Community Forum in the Threat Advisory board.