SALSA - The SAarbrücken Lexical Semantics Annotation and Analysis Project

Aims

The aim of SALSA is to create a semantically annotated corpus resource and to investigate methods for its utilisation.

Tasks

The construction of the corpus resource involves the following tasks:

• Creating a semantic annotation schema which allows for the integration of a wide range of phenomena
• Annotating the German 1.5M word TIGER corpus by hand with FrameNet semantic roles
• Concurrent development of a German "FrameNet light"
• Development of machine learning tools for supervised and unsupervised annotation of larger corpora

The semantically annotated corpus can be exploited on a whole range of different linguistic levels. Potential tasks are:

• Lexical semantics: the automatic acquisition of approximate or preferential meaning information for very large lexica
• Syntax: the improvement of statistical parsing techniques by training analysers on a combination of syntactic and semantic role information
• Intelligent search: the enhancement of linguistically guided techniques for information access and extraction

More details about SALSA I can be found on the Research page.