AMR preprocessing fails

[danielhers]

Could you please explain how to properly install the requirements for AMR preprocessing? I tried installing JAMR from GitHub but it fails. Did you install it from a different location?

[longxudou]

The original JAMR was installed in 2015 or 2016 in HPC, thus some packages were broken or not updated. The default setup script in JAMR repos is set to a version no longer available.

Today, I have installed JAMR again on my Macbook(Mojave 10.14.6) and uploaded new version in JAMR.

Here are the details:
First, please ensure the following env:
JAVA=8.0 (Download from oracle website)
sbt=1.0.2, scala=2.11.8 (Use SDKMAN! to install, like sdk install scala 2.12.1 )

Then, modify these files:

• jamr/project/build.properties:

sbt.version=1.0.2

• jamr/build.sbt:

// import AssemblyKeys._
// assemblySettings
name := "jamr"
version := "0.1-SNAPSHOT"
organization := "edu.cmu.lti.nlp"
scalaVersion := "2.11.8"
crossScalaVersions := Seq("2.11.8","2.11.12", "2.12.4","2.10.6")
libraryDependencies ++= Seq(
"com.jsuereth" %% "scala-arm" % "2.0",
"edu.stanford.nlp" % "stanford-corenlp" % "3.4.1",
"edu.stanford.nlp" % "stanford-corenlp" % "3.4.1" classifier "models",
"org.scala-lang.modules" %% "scala-parser-combinators" % "1.1.2",
"org.scala-lang.modules" %% "scala-pickling" % "0.10.1"
// "org.scala-lang" % "scala-swing" % "2.10.3"
)

• jamr/project/plugins.sbt:

resolvers += "Sonatype releases" at "https://oss.sonatype.org/content/repositories/releases"
resolvers += "Sonatype snapshots" at "https://oss.sonatype.org/content/repositories/snapshots/"
addSbtPlugin("com.eed3si9n" % "sbt-assembly" % "0.14.6")
// addSbtPlugin("com.github.mpeltonen" % "sbt-idea" % "1.6.0")
addSbtPlugin("com.typesafe.sbteclipse" % "sbteclipse-plugin" % "5.2.4")

[danielhers]

Thanks!

For JAMR to work,

• I think I had to run sbt before running ./setup in the JAMR directory.
• CDEC should be ${JAMR_HOME}/tools/cdec, right? It was failing due to Invalid range "\x{0CE6}-\x{0BEF}" in transliteration operator at ./tools/cdec/corpus/support/quote-norm.pl line 149., so I deleted line 149 there (it handles Tamil numerals).
• I also had to change scala-2.10/jamr-assembly-0.1-SNAPSHOT.jar to scala-2.11/jamr-assembly-0.1-SNAPSHOT.jar in the JAMR configs in scripts for the CLASSPATH to be right.

And for TAMR to work,

• I had to convert the GloVe file to word2vec format to make it loadable by Gensim: python -m gensim.scripts.glove2word2vec --input glove.840B.300d.txt --output glove.840B.300d.w2vformat.txt.
• I needed to run import nltk; nltk.download('stopwords') in Python.

[mhany90]

Installation now succeeds. However, on trying to run the 'amr_preproccess.sh' script I am getting the following error:

######## running tamr rule based aligner ...
WARN: use slow version of smatch api.
WordVec is loaded.
Traceback (most recent call last):
  File "/Users/mabdou/Desktop/phd/mr/mrc/parsers/HIT-SCIR-CoNLL2019/bash/../toolkit/tamr_aligner/rule_base_align.py", line 329, in <module>
    main()
  File "/Users/mabdou/Desktop/phd/mr/mrc/parsers/HIT-SCIR-CoNLL2019/bash/../toolkit/tamr_aligner/rule_base_align.py", line 323, in main
    align(opt)
  File "/Users/mabdou/Desktop/phd/mr/mrc/parsers/HIT-SCIR-CoNLL2019/bash/../toolkit/tamr_aligner/rule_base_align.py", line 120, in align
    graph = Alignment(block)
  File "/Users/mabdou/Desktop/phd/mr/mrc/parsers/HIT-SCIR-CoNLL2019/toolkit/tamr_aligner/amr/aligned.py", line 113, in __init__
    self.graph, self.attributes = self._get_graph(block)
  File "/Users/mabdou/Desktop/phd/mr/mrc/parsers/HIT-SCIR-CoNLL2019/toolkit/tamr_aligner/amr/aligned.py", line 251, in _get_graph
    for t in graph.triples():
**TypeError: 'list' object is not callable**
######## refreshing alignments ...
Traceback (most recent call last):
  File "/Users/mabdou/Desktop/phd/mr/mrc/parsers/HIT-SCIR-CoNLL2019/bash/../toolkit/tamr_aligner/refresh_alignments.py", line 47, in <module>
    main()
  File "/Users/mabdou/Desktop/phd/mr/mrc/parsers/HIT-SCIR-CoNLL2019/bash/../toolkit/tamr_aligner/refresh_alignments.py", line 18, in main
    assert len(lines) == 2
AssertionError
######## generating oracles ...
WARN: use slow version of smatch api.
######## adding extra mrp information ...
1
Traceback (most recent call last):
  File "/Users/mabdou/Desktop/phd/mr/mrc/parsers/HIT-SCIR-CoNLL2019/bash/../toolkit/amr_add_extra.py", line 39, in <module>
    amr_add_extra(args.input, args.output, args.extra)
  File "/Users/mabdou/Desktop/phd/mr/mrc/parsers/HIT-SCIR-CoNLL2019/bash/../toolkit/amr_add_extra.py", line 26, in amr_add_extra
    print(i, eid)
UnboundLocalError: local variable 'eid' referenced before assignment

[danielhers]

@mhany90 since Penman version 0.7.0, penman.Graph.triples is now a member variable instead of a method. So make sure to pip install 'penman<0.7'.