Notre approche respecte trois principes :
Building a syntax analyzer consists in moving from a language generation algorithm (context-free grammar) to a language recognition algorithm. For efficiency reasons, the recognition model (operator precedence, transition matrix, production language, etc) is not equivalent to the generation model. The transition of the first to the second consists in retaining only a certain type of information given by the grammar. This filtering corresponds to a loss of information. We then define the class of grammars for which the analyzer operates properly, that is the class of grammars that do not contain information eliminated a priori by the analyzer constructor. According to the quality of the information lost by different constructors, the corresponding grammar classes can be organized in an inclusion tree hierarchy. Unfortunately, one cannot give criteria so as to write a grammar of one or another class.
Or approach follows three principles::
\bibitem{Cousot-VMCAI03} P{.} Cousot. \newblock Un ana\-ly\-seur syn\-ta\-x\-ique pour gram\-mai\-res hors-con\-tex\-te as\-cen\-dant s{\'e}\-lec\-tif et g{\'e}\-n{\'e}\-ral. \newblock In \emph{Les tech\-ni\-ques de l'in\-for\-ma\-ti\-que}, Con\-gr{\`e}s AF\-CET 72, Bro\-chu\-re 1, Gre\-no\-ble, Fran\-ce, Nov.\ 1972. @InProceedings{Cousot72-1, author = {Cousot, P{.}}, title = {Un ana\-ly\-seur syn\-ta\-x\-ique pour gram\-mai\-res hors-con\-tex\-te as\-cen\-dant s{\'e}\-lec\-tif et g{\'e}\-n{\'e}\-ral}, pages = {391--405}, booktitle = {Les tech\-ni\-ques de l'in\-for\-ma\-ti\-que}, series = {Con\-gr{\`e}s AF\-CET 72, Bro\-chu\-re 1}, month = nov, year = 1972, address = {Gre\-no\-ble, Fran\-ce}, }
.
Last modified:
Friday, 04-May-2012 17:28:00 EDT