Microsoft Research IndiaEstablished January, 2005Seven research areas – Algorithms & Search – Cryptography, Security & Applied Math – Mobility, Networks, and Systems – Multilingual Systems – Rigorous Software Engineering – Technology for Emerging Markets – Vision, Graphics, and VisualizationContributions to Microsoft: – MultiPoint, Netra, Virtual India Image: Sridhar VedanthamCurrently ~55 full-time staff http://research.microsoft.com/indiaCollaborations with government,academia, industry, and NGOs in India
WikiBhasha: Machine Translation for Wikipedia A Kumaran & colleagues Multilingual Systems Group, MSR India ‘Bhasha’ in Sanskrit means ‘Language’
Wikipedia + MT = Win! Wikipedia Content by Language▪ Passionate Wikipedia communities▪ Highly skewed content in different languages▪ Use MT to help bootstrap Wikipedia into other tongues by bilingual Wikipedians▪ Corrections of MT used as data for improving translation algorithms
WikiBhasha: Functional View• As an edit layer on Wikipedia page CTF Linguistic & Collaborative Dictionary Services WikiBhasha UX Wikipedia API’s WikiBhasha User Wikipedia Community
WikiBhasha Beta• WikiBhasha announced in Oct 2010 – Open sourced (Apache & GPL2) – Released in www.WikiBhasha.org & MediaWiki SVN – Commitment to make all data as research resource
WikiBhasha and you!• Covered in 35+ languages across the world – Viral popularity
SpecNet: Shared wireless spectrum analysis for global coordinationKrishna Chintalapudi & colleaguesMobility, Networks & Systems, MSR India
Whitespace—Wasted resource dBm• Most of the allocated spectrum -110 -40 remains severely underutilized Mar 25 80% of TV spectrum remains unutilized 14:12 in rural areas of US 30% of the TV spectrum remains unutilized even in urban areas of US Under-utilization is expected to be Mar 24 much higher in developing nations 23:12 698 Freq in MHz 806 • White spaces Un-utilized spectrum: 100s of MHz! Prime spectrum: at sub GHz, coverage is 5-10 KM (compared to 100-250m for WiFi) FCC ruling Nov 14, 2008 allows opportunistic use of allocated spectrum FCC ruling Sep 23, 2010 allows use based on database OR sensing
Measuring spectrumCreation of a real-time spatio-temporal spectrumusage database• India is a large country 3200 Km• Information is not as readily available as in developed countries 3000 KmHow can we construct and maintain spatio-temporal spectrum usage maps?• A collaborative measurement platform is the key!• A network of spectrum sensing devices
SpecNet Spectrum Analyzer Remote User A first-of-its-kind platform that allows spectrum analyzers around theworld to be networked and efficiently used in a coordinated manner for spectrum measurement as well as implementation and evaluation of distributed sensing applications
The Power of SpecNet Enables remote real-time measurements • Help researchers to access real data from across Remote User Spectrum Analyzer the world to validate their modelsReal-Time Distributed Applications that UtilizesSpectrum Measurements• Researchers can implement and test their novel ideasusing the SpecNet API in real-timeConstruction and Maintenance of Real-Time WhiteSpace Spatio-Temporal Usage Maps• Can aid the operation of white space devices• Can help future white space service providers to plan theirinfrastructure deployments
Technology for Emerging MarketsResearch Group Goals Understand existing and potential technology users in developing communities Design and evaluate technology and systems that contribute to socio- economic development of poor and developing communities worldwide Collaborate with product groups, development-focused organizations, governments and industry for sustained, scaled impact Photo: Divya Ramachandran
CGNet Swara: Voice Portal for Citizen JournalismBill Thies, Technology for Emerging Markets, MSR Indiawith CGNet, MIT & International Center for Journalists
Voice Remains Primary Interfacefor Mobile Subscribers in India• Most subscribers lack smart phones Smart Phone: < 5% Feature Phone: 50-70% Basic Phone: 30-50% (e.g., music player)• Mobile Internet is rare, <3% of subscribers – Growing quickly, will not be free Source: McKinsey, IDC India
Dominant Interfaces: Voice, SMS • Voice calls remain central • Text messaging is common but not ubiquitous – 25% of subscribers send at least 1 SMS / day – 5% of subscribers check cricket scores on phone • Bluetooth is used by urban users – Peer-to-peer video sharing on feature phones Source: Mp Mobile Tracker
Literacy and Language Barriers Literate in EnglishNon-Literate
Literacy and Language Barriers Literate in English Literate in Other LanguageNon-Literate
Literacy and Language Barriers Literate in English Language Known Literate in Other LanguageNon-Literate
What will become the Web 2.0 forIndia’s 850 million mobile subscribers?
Journalism in Tribal Chhattisgarh Kui Number of speakers (inKurukh millions) Number of news outlets (print or audiovisual)Gondi 0 0.5 1 1.5 2 2.5 3• No news medium for tribal languages in India • Very few tribal journalists who know these languages • It is not legal to broadcast news over community radio in India!
CGNet SwaraVoice Portal for Citizen Journalism• Anyone can report news, issues, etc. in local language• Submissions are reviewed by moderators over the Web• Appropriate submissions are published: – For playback on audio channel – For browsing on web – Some submissions seed stories for posting on CGNet site + list
Application – Governance and Grievance Redress, News and Songs• About 40% of all posts concerns with governance – Government employment scheme (NREGA) workers not getting paid – Forest land not distributed under FRA – Situations of rural schools and health centers – Reports on corruption in public schemes – Teachers not being paid ▪ The teachers were paid their back wages one week after publication of the interview on CGNet Swara• Real time breaking news directly from the source in local languages – Reports on pollution and public health emergencies – Deaths due to cholera, diarrhea and malaria – Arsenic in water• Tribal stories and songs
http://www.cgnetswara.org/index.php?id=2847Father wanders for due NREGA wages, son dies in hospital... Manish Rai from Ambikapur says some days back I had heard an interview on CGnet Swara with a labourer called Pitbasu who had completed 100 days work in NREGA but had not been paid any wages. Today by chance I met him in the hospital and found that while Pitbasu was making rounds for his due NREGA wages his son died in the hospital. Is there any provision in NREGA to punish officials who has caused this grave incident? NREGA laws should be so strong that no one should wait for their wages as has happened with Pitbasu. For more on the story please contact Manish at 09826538904.