AHMEDABAD: Imagine a future Parliament where an MP from Gujarat can speak in Gujarati and it can be translated in Bengali, Tamil or Assamese real time for other members of the House.
A project helmed by the National Translation Mission (NTM) of Government of India is working on such a scenario, as Gandhinagar-based DAIICT has got grant of Rs 2 crore from the ministry of electronics & information technology for the period of three years to develop algorithm for Gujarati language. The primary domain for the projects will be e-governance, health and law.
Professor, Prasenjit Majumderwho is the principal investigator of information retrieval lab at DAIICT and who will head the project, said that it aims at artificial intelligence (AI)-based machine translation of Gujarati and other scheduled languages of India.
“Machines cannot replace humans completely when it comes to understanding or translating languages faithfully. However, our goal is to ensure that the share of intervention in total translation will be not more than 20% with good quality output,” he said.
Giving a peek into the process of ‘teaching’ computers in Gujarati, he said that majority of the AI and machine learning (ML) projects depend on English. “The syntax of English is different from the majority of Indian languages – the latter is identified as free-form syntax, as it’s not bound by subject-verb-object formation of English,” he said.
Thus, when it comes to Indian languages, the researchers or translators cannot simply lift the available software, as it may not give desired results. The DAIICT experts said that computerization and need for computers in local languages such as voice-assisted navigation or translation of any text has created need for developing a unique logic and semantics.
“This is where the work of information technology is required – our team would develop algorithms that can understand the complexity of a language, formulate a logic to process any input, and provide relatively cleaner results,” said Prof Majumder. “Especially post pandemic when we are looking at a large cache of knowledge required to reach the last person, such tools can be handy.”
A project helmed by the National Translation Mission (NTM) of Government of India is working on such a scenario, as Gandhinagar-based DAIICT has got grant of Rs 2 crore from the ministry of electronics & information technology for the period of three years to develop algorithm for Gujarati language. The primary domain for the projects will be e-governance, health and law.
Professor, Prasenjit Majumderwho is the principal investigator of information retrieval lab at DAIICT and who will head the project, said that it aims at artificial intelligence (AI)-based machine translation of Gujarati and other scheduled languages of India.
“Machines cannot replace humans completely when it comes to understanding or translating languages faithfully. However, our goal is to ensure that the share of intervention in total translation will be not more than 20% with good quality output,” he said.
Giving a peek into the process of ‘teaching’ computers in Gujarati, he said that majority of the AI and machine learning (ML) projects depend on English. “The syntax of English is different from the majority of Indian languages – the latter is identified as free-form syntax, as it’s not bound by subject-verb-object formation of English,” he said.
Thus, when it comes to Indian languages, the researchers or translators cannot simply lift the available software, as it may not give desired results. The DAIICT experts said that computerization and need for computers in local languages such as voice-assisted navigation or translation of any text has created need for developing a unique logic and semantics.
“This is where the work of information technology is required – our team would develop algorithms that can understand the complexity of a language, formulate a logic to process any input, and provide relatively cleaner results,” said Prof Majumder. “Especially post pandemic when we are looking at a large cache of knowledge required to reach the last person, such tools can be handy.”
,