资 源 简 介
First described in the paper "Probabilistic Frame-Semantic Parsing" published at the NAACL 2010 conference, and improved in three papers published at ACL 2011, NAACL 2012 and *SEM 2012, this project uses the theory of frame semantics (http://framenet.icsi.berkeley.edu) and statistical machine learning to produce shallow semantic structures from raw natural language text.
An example frame-semantic parse of a sentence is shown below:
Each row under the sentence correponds to a semantic frame and its set of corresponding arguments. Thick lines indicate targets that evoke frames; thin solid/dotted lines with labels indicate arguments. N_m under “bells” is short for the Noise_maker role of the NOISE_MAKERS frame.
So far, the SEMAFOR parser has been tested for English, and uses the FrameNet 1.5 lexicon as a reference for analyzing te
