A karaka based approach to parsing of Indian languages
Akshar Bharati, Rajeev Sangal · 1990
A karaka based approach for parsing of Indian languages is described. It has been used for building a parser of Hindi for a prototype Machine Translation system.A lexicalised grammar formalism has been developed that allows constraints to be specified between 'demand' and 'source' words (e.g., between verb and its karaka roles). The parser has two important novel features: (i) It has a local word grouping phase in which word groups are formed using 'local' information only. They are formed based on finite state machine specifications thus resulting in a fast grouper. (ii) The parser is a general constraint solver. It first transforms the constraints to an integer programming problem and then solves it.