Designing a Bangla parser using non-deterministic push down automata

Abstract

The goal of language processing is to make machines be able to read human language comprehensively and use human language to communicate with human beings. To achieve this goal, the first step is to parse the sentence structure correctly. In this paper, we propose a parsing technique for Bangla Language based on Non-Deterministic Push-Down Automata (NPDA). The NPDA parser takes the Context Free Grammars (CFG) of Bangla Language for preprocessing. The NPDA is efficient especially when large number of CFGs needs to be processed. The predictive parser needs to generate parse table from the CFG if the number of CFGs are large then it is difficult to construct such a big parse table. The NPDA does not require constructing any parse table and it can process the CFG directly. The parser can detect Bangla sentences of all forms whether they are syntactically and grammatically correct or not. Finally, we compare these two parsers with their complexity and efficiency issues. Comparing with these important issues, the NPDA parser gives the better result than predictive parser. Sufficient examples and figures are described to explain the parsing idea.