KURD

In its origin, KURD is a formalism for shallow post-morphological processing as described in two papers presented on  NeMLAP/CoNLL'98  and  NLPRS'97 . He works in a number of IAI  projects: MULTILINT, MULTIDOC, TETRIS and the DUDEN-project mainly on grammar checking and style control. KURD usually expects input from the morphological analyzer MPRO. This input is modified according to a number of rules. KURD-rules are specified in a rule-file and sequentially  applied to objects read from the input stream. In the DUDEN-grammar there are now more than 1000 rules for detecting errors in German texts. A less ambitious grammar exists also for English.

With a new structure building operator, KURD, however, goes beyond shallow processing. This page provides an overview of the data structure in KURD, its rule format and options.

Formal description of KURD

Publications