A context sensitive tabular parsing algorithm
Frank Hadlock · 1990
Tlie parsing algorithm due to Younger is applica ble to any context free language but requires that the grammar be in Chomsky normal form.Resides es tablishing an 0 (n'*) upper bound to time complex ity needed to parse context free languages, Younger's algorithm has the advantages that it docs not re quire the grammar to be unambiguous and that it degrades gracefully while parsing "ungrammatical" strings.These are advantages desirable in processing nnlurnl language, if not coupled with the restriction that the grammar be in Chomsky normal form.In this article, a general tabular algorithm is de veloped which uses the tabular structure employed by Younger's algorithm.Instead of nonterminal charac ters (or syntactic types) as table entries, a construct is used which is similar to the LR(0) items employed by Knulh's LR(k) parsing algorithm, and by the Gra ham, Harrison, R u eeo algorithm.Graham, Harrison and Russo use these items as table entries, obviating the need to employ Chomsky normal form.The algo rithm presented in this article employs such items for not only the replacement phrase, but also the left and right context strings in context sensitive rules.The resulting algorithm is applicable to context sensitive grammars without erasure rules and is suitable as a platform for semantic case frame analysis of natural language. 1In tro d u ctio nFor basic concepts in phrase structure languages and grammars, as well as the parsing algorithms referenced in this article, the reader is referred to [6].