The following works are class projects developed 1998-2001 during my time at University of Potsdam.
CyLaRP: A unification-based parser
CyLaRP is an efficient parser for unification-based grammar formalisms. Its foundations were defined by M. Tomita (1985). CyLaRP grammars are defined in a language in the XML framework. It is implemented in C++ and fully portable. Development has been supported by Agentscape AG, Berlin
Deep linguistic analysis of input sentences; parsing using an augmented Earley Chart Parser (see below) and a unification-based grammar. Semantic analysis with Lambda-DRT. Knowledge Base in LOOM. My work on that project concentrated on semantic construction (together with R. Ludwig), the Web interface, the LOOM model checking interface and the parser.
A more precise overview in English of this ongoing project is available.
The algorithm is based on Earley's 1970 algorithm. An informal description is given in the documentation. The implementation was done in SICSTUS PROLOG, the feature structures are represented using GULP-3 (Covington 1994). Documentation is in German language - sorry.
Download earley-fs-parser.zip (272 KB)
Verbs taken from the semantic field of cooking (in English and German language) were coded using the (rather traditional) knowledge representation system LOOM. An external module (Lexical Option Finder, Stede 1993) can find the right verbs for semantically defined situations.
Download cook1099.zip (219 KB)
The document gives an introduction in Kamp's DRT and its implementation in Prolog. Construction is done using the Lambda-Calculus (as invented in Stuttgart and Saarbr¸cken). The document summarizes the works of [Kamp/Reyle 1993] and [Blackburn/Bos 1999]. The documen is written in German - sorry for that.
Download lambda_drs.zip (87 KB)
Handout Revision-Based NLG: reitter_robin_revisionbasednlg_handout.pdf (87 KB)
