Offering improved services, Google Maps has rolled out automatic transliteration into 10 Indian languages which are Hindi, Gujarati, Kannada, Bangla, Marathi, Malayalam, Punjabi, Odia, Tamil, and Telugu. It is to note that transliteration is not translation. When a sentence is transliterated, the same words are written in a different script that will enable users to have better search results in their native languages.
In order to bring more accurate results, Google said that it has built an ensemble of learning models that will transliterate names of Latin script point of interest (POI) areas within the country. People living in the country who are not well versed with English, can use this new automatic transliteration technology that Google has introduced.
Prior to the introduction of this improved feature, the transliteration of native script used by people did not work on the algorithm (that company has now introduced) and therefore, the search results that popped were not as accurate as they needed to be. According to the company’s blog post, the new addition of learning models has enabled Hindi transliteration to improve by 3.2 times as far as coverage is concerned and 1.8 time quality improvement was seen. Similarly, with the transliteration in the Bengali native language, coverage improved 19 times and quality improved 3.3 times. Odia language transliteration witnessed a massive improvement of 960 times in terms of coverage, thanks to a new ensemble of learned models.
The company noted that the new ensemble of learned models along with various transliteration dictionaries as well as a module for acronyms has significantly increased the quality and coverage for some languages.
“As with any machine learned system, the resulting automatic transliterations may contain a few errors or infelicities, but the large increase in coverage in these widely spoken languages marks a substantial expansion of the accessibility of information within Google Maps in India,” Google said in its blog.