Share on Facebook Tweet on Twitter Share on LinkedIn Share by email
Vikram Dendi

Vikram Dendi
TECHNICAL & STRATEGY ADVISOR
.

Vikram Dendi is the Strategy Director for Microsoft Research and acts as the Technical & Strategy Advisor to the Head of Microsoft Research. He is responsible for increasing the impact of MSR’s research investments and also currently oversees the teams supporting technology transfer, IP strategy and technical operations.

Vikram has been at Microsoft Research for five years and his prior responsibilities included helping transform research technologies like machine translation into high value developer and user experiences. He was responsible for shaping the product direction and business strategy for Microsoft’s translation platform, which is used by most major Microsoft products, including Bing, Office and Windows, as well as many partner products (Facebook, eBay, Twitter, etc.). Over the course of his tenure at Microsoft, he contributed to a number of other “new product” efforts in a variety of areas, including Maps, Location Based Services, Phone, and Emerging Markets.

Prior to Microsoft, he worked for Real Networks where he played a key role in establishing the company’s open source strategy and was responsible for designing, building, and shipping consumer products for the Mac and Linux platforms.

Vikram graduated with honors from California Institute of Technology and while an Entrepreneurial Fellow in a National Science Foundation sponsored program, he founded his first startup company in the area of tablet computing. Vikram's research interests include Computer Human Interaction, Social Computing, Intelligent Interfaces and Software Agents.

You can connect with him here and follow him @vikman.

Publications
What's New
Microsoft Translator Now Supports 50 Languages with the Additions of Bosnian, Croatian, and Serbian
Sarajevo, Bosnia and Herzegovina Zagreb, Croatia Belgrade, Serbia   We are happy to announce that have added three new languages– Bosnian, Croatian, and Serbian to our list of supported languages. This brings the total number of languages supported by Microsoft Translator to 50! Bosnian, Croatian, and Serbian are welcome additions to the list of Microsoft Translator’s supported languages. Although similar in form and structure, these languages belong to countries that have a unique histories and cultures. These languages are primarily spoken in South East Europe, in the region between Italy and Greece. The nations of Bosnia and Herzegovina, Croatia, and Serbia are major tourist destinations, with millions of people visiting each year. The capital of Bosnia and Herzegovina, Sarajevo, is well known for its old bazaar known as Bascarsija, which was built in the 15th century. Croatia’s coastline draws hordes of visitors to its shores, and the capital of Serbia, Belgrade, has a booming nightlife. These three nations also host a variety of business and professional opportunities within their borders. We are excited to launch these translation systems, making it possible for people around the globe to have a richer understanding of these cultures and this area of the world. Microsoft Translator’s automatic translation engine makes instant translation a reality no matter where you are. We offer a wide variety of languages spoken all over the world, and beyond— from Africa to Asia, North America to Europe...and even Klingon. Whether you use Translator via Bing.com/Translator, on your desktop using the Windows 8 app, in Office*, through Skype Translator, SharePoint, Lync, in your own app or service through the Translator API, or simply through your favorite social media or consumer review app, you will now be able to communicate in 50 languages and have your world in your words. You can even check out these new translation systems on the go using the Translator app for Windows Phone, featuring camera mode** for translating text on walls, signs, et cetera-- wherever they may be! Learn More about how Microsoft Translator: Microsoft Translator on YouTube How Microsoft Translator works Try it out on your mobile device: Windows Phone, or through our partners at Translate.com for Android or iOS * Newly released language will only appear in Microsoft Office applications if you update the service from within the app. See the steps here: http://aka.ms/TranslatorAddNewLanguageInOffice ** Serbian only support for the moment.
Translator Solutions in Action: Internet Explorer
  Team: Internet Explorer Solution: Customer Support The Internet Explorer (IE) team at Microsoft uses the Microsoft Translator API to translate non-English customer reports from around the world, allowing them to resolve customer issues faster and more effectively. The API has also improved IE’s ability to triage customer comments and conduct data mining on text that is not written in English. The Translator API was quick and painless for IE to incorporate into their current projects and workflows, and Translator’s marketplace portal has been easy to use— allowing them to effortlessly monitor and change their usage quotas.  
Simplicity is Possible in a Multi-Lingual, Omni Channel Support World
The following is a guest post by the Microsoft Translator Partner, Lionbridge Technologies, who developed GeoFluent as a solution to address the challenge of real-time translation of user generated content leveraging the Microsoft Translator automatic translation service and customization capabilities of the Translator Hub. Let’s face it: customers appreciate simplicity. Nothing saves an angry customer from becoming an ex-customer like simple, seamless customer support. Savvy businesses offer up to a dozen contact channels to deliver support at the right time and place for a customer, but many are stymied by the complexity of providing this level of support for their entire customer base, and in a variety of different selling mediums. Consider the multi-lingual nature of North America alone. According to a recent survey, 30% of North Americans do not consider the English language as their native language. The number of limited English proficiency (LEP) individuals in the United States has also grown by 81 percent since 1990. Finally, nearly one in ten working-age U.S. adults—19.2 million persons aged 16 to 64—are considered limited English proficient. The complexity, customer service executives would say, is that they cannot possibly staff for support of all the languages of all their customers any given point. Another factor making effective customer service seemingly complex is the dynamic way customers can interact with service providers. This interaction is omni-channel, whereby customers can not only buy products from selling organizations, but also reverse direction and interact and provide feedback to that sell for all to see. This has put new pressure on sellers to quickly and effectively manage this interaction, or risk a hostile reputation. Indeed, according to Forrester Research, 67% of today’s Internet users would prefer to find (pre- and post- sales) answers online. Modern, simple customer support is dynamic – whether one to one contact channels like email ticketing and chat, communication takes place in near real time. Chat, for example, can be an effective way for customers to reach out to customer service representatives to have their issues resolved, and learn about new products. At the same time, technology has made it possible to allow customers to converse in their own native language, and provide the same capability for company representatives and other customer service staff. This is achieved by enhanced, personalized machine translation. As a result, multinational businesses – or even businesses with multilingual customer bases – can significantly broaden their reach, boost brand loyalty and cost-effectively support customers, regardless of language, location or device. Best of all, translation processes can be specially mindful of industry terms that need to be carefully translated, security concerns, and easy accessibility by both agent and consumer. Enhanced, personalized machine translation is definitely not as perfect as human translation. But that’s okay. What it does do is aide chat conversations that cannot wait for long-term translation perfection to be actionable, understandable, and immediate. When deploying machine translation in a support environment, organizations must consider: How does the solution integrate into existing software/workflows The preservation of brand/product terms. How to maximize translation quality through translation engine customization and agent/moderator training? Does the solution offer adequate safeguards for the handling of PII. This is the heart of what GeoFluent by Lionbridge does every day for all of our customers. As a result of our partnership with Microsoft Translator, we can help answer these challenges and make the customer service translation process simple, actionable, understandable, and immediate. Your customer base is only growing more diverse and channel-savvy. Providing simple, cost effective in language customer support is within your grasp. Please visit http://geofluent.lionbridge.com/ to learn how state of the art machine translation can make multilingual, multi-channel, customer support simple. By: Greg Belkin, Director of Product Marketing and Product Management, Lionbridge.
Microsoft Translator Introduces Yucatec Maya and Querétaro Otomi for UNESCO's International Mother Language Day
In our ongoing effort to empower language communities to preserve their languages and cultures, we are excited to introduce Yucatec Maya and Querétaro Otomi to Microsoft Translator’s ever-growing list of supported languages. These language systems were developed in collaboration with community partners in Mexico, who created the automatic translations systems to permanently bridge the translation gap between these endangered languages and the rest of the world. The systems themselves were built using the Microsoft Translator Hub, a Translator product which is available for free to allow any group to create its own unique translation systems. Photo courtesy of Instituto Querestano de la Cultura y las Artes Using the Hub, our community partners took important steps to preserve their language and culture. The Yucatec Maya translation system was built by the Universidad Intercultural Maya de Quintana Roo (UIMQROO), a university in the southwestern Mexican state of Quintana Roo that was created to provide higher education to the Maya population of the region. Native to the Yucatan region of Mexico and Belize, Yucatec Maya is spoken by fewer than 800,000 people, with less than 59,000 monolingual speakers. The language is descended from the language of the ancient Mayan empire, which is well-known for its art, architecture, as well as its expertise in astronomy. The Querétaro Otomi language system was created by the Instituto Queretano de la Cultura y las Artes (IQCA), an institute in western central Mexico whose mission is to encourage artistic and cultural development and to promote equity and equality of opportunity within the State of Querétaro. Querétaro Otomi is an endangered language from the region that is only spoken by 33,000 people and has fewer than 2,000 monolingual speakers. The release of Maya and Otomi helps to celebrate the UNESCO’s International Mother Language Day, an annual international event which aims “to promote the preservation and protection of all languages used by peoples of the world.” According to UNESCO, “if nothing is done, half of 6,000-plus languages spoken today will disappear by the end of this century.” Maya and Otomi are indigenous languages from Mexico which are both currently threatened. Although they are still in use, the number of speakers is decreasing and younger people are not speaking them as actively as their elders. The new automatic translation systems will help the Maya and Otomi people safeguard their language and culture for generations to come. Over the years, Translator has worked closely with a variety of language community partners to encourage language preservation and, through it, intercultural communication. In the past, these community partners have used the Hub to create translation systems for languages such as Hmong Daw, Welsh, and Urdu. The Hub allows organizations such as UIMQROO and IQCA to leverage the computing power of Microsoft Translator’s machine-learning back end as well as its existing translation models to create unique and customized translation systems. The Translator Hub is a powerful tool for organizations that have specific translation needs, such as language preservation. It also allows organizations to create domain-specific systems, including industry-specific translation systems (for instance, for the medical or financial sectors) and business-specific systems that are customized to the company’s internal style and terminology. In addition to the Hub, Translator also supports a wide variety of products to connect individuals across language barriers, including the Translator API, which can be used to translate web pages and apps in real time into 45+ languages, as well as powering the translation features in the Microsoft Office suite of products. Most recently, Microsoft Translator and Skype introduced Skype Translator, a next–generation speech-to-speech translation platform which allows users to converse in different languages in near-real time. To learn more about International Mother Language Day, and what Microsoft is doing to support technology on this front, please visit the Official Microsoft Blog. Learn More about the Translator Hub and Language Preservation: Microsoft Translator Hub on www.Microsoft.com/Translator Translator Hub YouTube playlist Microsoft Translator Community Partners
Translator Solutions in Action: Bing
  Team: Bing Solution: Customer Support, Website Localization The Bing search engine offers its services around the globe, with millions of requests entered in languages other than English. In order to improve foreign language search results, the Whole Page Relevance team at Bing needed a way to translate millions of phrases from more than 10 graphic device interface languages into English. Rather than relying on a costly and time-consuming third-party human translator service, the team was able to use the Microsoft Translator API to quickly and efficiently translate millions of documents into English. This allowed the platform to better understand foreign language search requests and produce higher quality results.