Shalmaneser: A Shallow Semantic Parser
March 13, 2007: Release 1.1
- Bugs removed
- Support for TreeTagger for English POS tagging
- English pre-trained classifiers for FrameNet 1.3
parser: sentence.c:112: convert_sentence: Assertion `0' failed.This error is not in Shalmaneser itself, but in the collaboration between POS tagger and parser. Apparently, newer versions of TreeTagger give more detailed POS tags (like IN/that instead of IN as POS tag for that), and the Collins parser can't deal with that. If you see this error, processing will have failed. Please Have a look at the file Pkg/TreetaggerInterface.rb, Method convert_to_collins(line). If it doesn't look like this, replace the existing code with this:
def convert_to_collins(line) line.chomp! # if there are any slashes in the tag, just ignore them (and everything after them) if line =~ /^(.+)\// line = $1 end return line.gsub(/^PP/,"PRP").gsub(/^NP/,"NNP").gsub(/^VV/,"VB").gsub(/^VH/,"VB").gsub(/^SENT/,".+") end
Shalmaneser is a supervised learning toolbox for shallow semantic parsing, i.e. the automatic assignment of semantic classes and roles to text. The system was developed for Frame Semantics; thus we use Frame Semantics terminology and call the classes frames and the roles frame elements. However, the architecture is reasonably general, and with a certain amount of adaption, Shalmaneser should be usable for other paradigms (e.g., PropBank roles) as well. Shalmaneser caters both for end users, and for researchers.
For end users, we provide a simple end user mode which can simply apply the pre-trained classifiers for English (FrameNet annotation / Collins parser) and German (SALSA Frame annotation / Sleepy parser). For researchers interested in investigating shallow semanticparsing, our system is extensively configurable and extendable.