huak95/amr-reader

Abstract Meaning Representation (AMR) reader w/ gradio

Quickstart

  python amrreader/main.py [-h] [-g GRAPH] [-n] [-p] [-e] [-v VISUALIZATION] indir outdir
  e.g., python amrreader/main.py amrreader/test/test_amr_doc/ output/ -v=n


  positional arguments:
    indir                 directory to AMR input files
    outdir                output directory

  optional arguments:
    -g GRAPH, --graph GRAPH
                          generate AMR graphs -g=n: standard graphs
                                              -g=s: simplified graphs
    -n, --node            generate AMR nodes
    -p, --path            generate AMR paths
    -e, --entity          generate named entities
    -v VISUALIZATION, --visualization VISUALIZATION
                          generate html visualization -v=n: standard graphs
                                                      -v=s: simplified graphs

• Ptyhon3
• Your input should be raw AMR format (see amr-reader/tests/test_amr_doc/test).
• If you would like to use AMR visualization functionality, please install PyGraphviz first.
  

Example

  # ::snt I am cautiously anticipating the GOP nominee in 2012 not to be Mitt Romney.

  (a / anticipate-01
  :ARG0 (i / i)
  :ARG1 (n / nominate-01 :polarity -
        :ARG0 (p2 / political-party 
                    :wiki "Republican_Party_(United_States)" 
                    :name (n3 / name :op1 "GOP"))
        :ARG1 (p / person
              :wiki "Mitt_Romney" 
              :name (n2 / name :op1 "Mitt" :op2 "Romney"))
        :time (d / date-entity :year 2012)))

Visualization:

  Green Ellispe: concept
  Orange Ellispe: predict with sense
  Black Ellispe: constant
  Blue Rectangle: named entity

Citation

If you would like to cite this work, please cite the following publication:
Unsupervised Entity Linking with Abstract Meaning Representation.

Demo

AMR based Entity Linker

API

AMR based Entity Linker