At the LREC workshop New Challenges for NLP Frameworks [3] we released a new component for GATE: The Predicate-Argument Extractor (PAX) [4].
The idea of the PAX component is to facilitate experiments with different parsers (MiniPar, RASP, Stanford, SUPPLE, and an NP chunker) within an NLP pipeline. While GATE makes it technically easy to swap out one parser for another, each of them has a different output format, which would require adapting downstream components. PAX extracts normalized predicate-argument structures (PASs) (predicate - subject - object) from the output from all parsers.
For more details and the (open source) download, have a look at the PAX Webpage [4], as well as our research paper Predicate-Argument EXtractor (PAX) [5].
