TOWARDS A RULE PREFERENCE BASED APPROACH TO LSD PARSING

Natural language parsing plays an import rule in the machine translation research. This paper first introduces some fundamental concepts with LSD based syntactic parsing, then proposes a rule preference based LSD parsing algorithm. In this approach, both the preference based on structural information and the preference based on lexical statistical information is considered in syntactic disambiguation. The complexity of the algorithm is given. The results show that this approach can improve the accuracy of the parsing accuracy and recall of properly used LSF by around 10% under the condition that the parser’s performance unchanged. This approach shows itself a promising technique in efficient natural language parsing.