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.
- A Kumaran, Naren Datha, Vikram Dendi, and Ashwani Sharma, WikiBhasha: OurExperiences with Multilingual Content Creation Tool for Wikipedia, in Proceedings of the Wikipedia India Conference 2011, Wikimedia Foundation, December 2011.
- kumarana, narend, Ashwani Sharma, and Vikram Dendi, WikiBhasha:OurExperiences with Multilingual Content Creation Tool for Wikipedia, in Proceedings of Wikipedia Conference India, Wikimedia Foundation, November 2011.
- A Kumaran, Naren Datha, B Ashok, K Saravanan, Anil Ande, Ashwani Sharma, Sridhar Vedantham, Vidya Natampally, Vikram Dendi, and Sandor Maurice, WikiBABEL: A System for Multilingual Wikipedia Content, in in Proceedings of the 'Collaborative Translation: technology, crowdsourcing, and the translator perspective' Workshop (co-located with AMTA 2010 Conference), Denver, Colorado, Association for Machine Translation in the Americas, 31 October 2010.
- A Kumaran, Naren Datha, K Saravanan, Vikram Dendi, and Sandor Maurice, WikiBABEL: A Wiki-style Platform for Creation of Parallel Data, in the 47th Annual Meeting of the Association for Computational Linguistics and the 4th International Joint Conference on Natural Language Processing of the Asian Federation of Natural Language Processing (ACL/IJCNLP-2009), Singapore, Singapore, Association for Computational Linguistics, August 2009.
- Vikram Dendi, The Emergence Of Machine Translation, in MSDN Magazine, 1 January 2009.
|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.
Tue, 03 Mar 2015 17:00:00 GMT
|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
Mon, 23 Feb 2015 14:00:00 GMT
|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.
Wed, 18 Feb 2015 18:17:00 GMT
|Translator Solutions in Action: Microsoft Services Global Delivery
Over the next few weeks, we'll be showcasing how various teams around Microsoft have been able to use Translator to improve their internal and external operations in areas such as readiness, communications, customer support, forums and user groups, and web localization. Translator has proved to be a valuable tool for many teams across Microsoft, and we're happy to be able to share their stories. Team: Microsoft Services Global Delivery Solution: Customer Support Microsoft Services Global Delivery (MSGD) serves customers around the globe, especially catering to markets such as Asia-Pacific (including China) and Latin America. For overseas engagement in these regions, MSDG needs quick document translation - but the volume is too high for human translators. By using the Microsoft Translator API, MSGD was able to develop internal translation tools which can instantly read any Microsoft Office or PDF document and translate it in-place, using secure connections, without compromising any content or images. They also created a system which allows customers to communicate with them through two-way translation of XML, bridging the language gap when dealing with their customers' legacy code. MSGD has been using their new translation system since 2013 for documentation in the Asia-Pacific and Latin American markets. The system has increased their business agility, allowing for quicker turnaround and seamless collaboration with their customers.
Sat, 07 Feb 2015 01:34:00 GMT
|WorldWideScience.org Provides Multi-Language Access to the World's Scientific Research
How do you make libraries of billions of pages of scientific research, published in multiple languages, accessible to people around the globe? This was the problem faced by the WorldWideScience Alliance, a multinational partnership whose mission is to eliminate barriers in finding and sharing research across national boundaries. To solve this problem, WorldWideScience reached out to Deep Web Technologies which specializes in multilingual, federated search solutions across multiple industries. The result of this partnership was the WorldWideScience.org web portal that can search 100 different databases across 70 different countries, and then rank and translate the results into the user's preferred language. Millions of scientific articles are published around the world each year in wide variety of languages, but only some of them are available on the worldwide web using conventional search engines. Scientific publications are typically found in what is called the "Deep Web", which consists of documents, images, and records located in an often unconnected series databases throughout the globe. Using Microsoft Translator, Deep Web Technologies was able to create WorldWideScience.org — a consolidated portal, accessible through the web, which is able to search these worldwide databases and translate the results into 10 different languages. Users can choose to view scientific papers, multimedia, or research data on their desktop computers or mobile devices. The Microsoft Translator API provided Deep Web Technologies with the high level of scalability and reliability required for the project, and the translation API was easily integrated with the wide variety of data sources WorldWideScience.org pulls from. The completed portal improves global access to scientific research, encourages international collaboration, and provides new opportunities to share data. WorldWideScience.org now increases access to scientific and technological research worldwide, facilitates international collaboration, and provides new opportunities for research in multiple industries. The site handles approximately 70,000 queries and 1 million page views each month, and all traffic, including that from automated crawlers and search engines, amounts to approximately 70 million transactions per year. The multilingual, federated search solution implemented by WorldWideScience Alliance and Deep Web Technologies is applicable across a wide range of industries, and could be used for solutions ranging from customer support to organizational readiness. To learn more about WorldWideScience.org and multilingual, federated search technology, read the full case study.
Thu, 15 Jan 2015 17:30:00 GMT