India is home to the second highest number of internet users in the world. However, a vast majority of the country’s population interacts with the web using local languages, instead of English. To make Google Maps more accessible to users in India, Google has created an ensemble of learned models to transliterate names of Latin script POIs into 10 popular languages spoken in the country: Bangla, Gujarati, Hindi, Kannada, Malayalam, Marathi, Odia, Punjabi, Tamil, and Telugu.
Until now, names of nearly all major Indian places of interest (POIs) in Google Maps were only available in the Latin script. These names were also often combined with acronyms based on the Latin script, as well as some Indian language words.
Explaining the problem with the mixed-language representations, Cibu Johny, Software Engineer at Google Research and Saumya Dalal, Product Manager at Google Geo, wrote in a blog post:
Consider a user in Ahmedabad, Gujarat, who is looking for a nearby hospital, KD Hospital. They issue the search query, કેડી હોસ્પિટલ, in the native script of Gujarati, the 6th most widely spoken language in India. Here, કેડી (“kay-dee”) is the sounding out of the acronym KD, and હોસ્પિટલ is “hospital”. In this search, Google Maps knows to look for hospitals, but it doesn’t understand that કેડી is KD, hence it finds another hospital, CIMS. As a consequence of the relative sparsity of names available in the Gujarati script for places of interest (POIs) in India, instead of their desired result, the user is shown a result that is further away.
Google says it has added names in the 10 languages to “millions” of POIs in India and increased the coverage twenty-fold in some languages. Millions of Indian users who don’t speak English can now easily find hospitals, grocery stores, train stations, and other places of interest in their own language in Google Maps. In the future, Google plans to use the ensemble for transliteration of other entities within Maps and extend it to other popular Indian languages and scripts.
We may earn a commission for purchases using our links. Learn more.