Hyderabad: Researchers at the International Institute of Information Technology, Hyderabad (IIIT-H), have unveiled India’s first deep Tamil syntactic parser. Spearheaded by Prof. Parameswari Krishnamurthy, the project received funding from the Tamil Virtual Academy, under the information technology and digital services department of the government of Tamil Nadu.
The parser, named Kalaignar Tamil parser in homage to former Tamil Nadu chief minister M. Karunanidhi, has been developed as part of a TN government-funded initiative aimed at bridging language gaps and advancing machine translation systems. The project involved extensive human annotation of over 100,000 Tamil words, forming the backbone of the parser's algorithm.
Prof. Krishnamurthy spoke about the pivotal role of parsers in facilitating machine understanding of textual content, essential for accurate translation. The syntactic parser, a key component of the project, enables the breakdown of sentences into constituent parts, including parts of speech and sentence structure, laying the groundwork for semantic analysis.
The IIIT-H team’s parser, built on open collaboration principles, leverages tree banking, and universal dependency features from a vast community effort spanning over 100 languages. Beyond machine translation, the researchers envision applications in language teaching and testing, with plans to integrate the tool into assistive language learning apps.
The Kalaignar Tamil parser was unveiled during the KaniTamil24 international conference, organised by the Tamil Nadu government to promote language technology. Dr Palanivel Thiyagarajan, the Tamil Nadu IT minister, formally launched the tool, underscoring its significance in advancing Tamil language computing.
In addition to the parser unveiling, IIIT-H researchers presented a paper on domain adaptation of bi-directional neural machine translation systems involving Tamil to Telugu. A panel discussion at the conference delved into the ethical considerations surrounding AI in linguistics, advocating for a regulated and conscientious approach.
