Created
September 25, 2014 20:26
-
-
Save dimazest/33ee3de4fe219c01576a to your computer and use it in GitHub Desktop.
Produce playground
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
[] | |
bin = ../../bin | |
corpora = %{bin}/corpora | |
fowler_corpora_py = %{bin}/fowler.corpora-py | |
bnc_corpus = bnc+ccg://${PWD}/CCG_BNC_v1 | |
# Get the wordsim353 similarity dataset | |
[data/wordsim353.csv] | |
recipe = | |
mkdir -p data | |
# Lowercase the words and replace "troops" with its stem "troop" | |
# This transformation is needed because word sems will be used to extract co-occurences. | |
curl -s https://bitbucket.org/dimazest/phd-buildout/raw/tip/notebooks/downloads/wordsim353/combined.csv \ | |
| tr '[:upper:]' '[:lower:]' | sed -e 's/troops/troop/g' \ | |
> %{target} | |
# Get the Rubenstein and Goodenough 65 similarity dataset | |
[data/rg65.csv] | |
recipe = | |
mkdir -p data | |
curl -s https://bitbucket.org/dimazest/phd-buildout/raw/tip/notebooks/downloads/RubensteinGoodenough/EN-RG-65.txt \ | |
> %{target} | |
# Extract all the words from the wordsim353 similarity dataset | |
[out/wordsim353_targets.csv] | |
dep.input = data/wordsim353.csv | |
recipe = | |
mkdir -p out | |
# Get the first colum | |
cut %{input} -d, -f 1 > t | |
# Append the second column | |
cut %{input} -d, -f 2 >> t | |
# The header | |
echo ngram > %{target} | |
# Get rid of duplicates and the column names ("word 1", "word 2") | |
cat t | sort | uniq | grep -v 'word 1' | grep -v 'word 2'>> %{target} | |
rm t | |
# Extract all the words from the Rubenstein and Goodenough 65 similarity dataset | |
[out/rg65_targets.csv] | |
dep.input = data/rg65.csv | |
recipe = | |
mkdir -p out | |
# Get the first colum | |
cut %{input} -f 1 > t | |
# Append the second column | |
cut %{input} -f 2 >> t | |
# The header | |
echo ngram > %{target} | |
cat t | sort | uniq >> %{target} | |
rm t | |
# Count the word frequencies, POS tagged | |
[out/%{experiment}_dictionary-bnc-pos.h5] | |
dep.targets = out/%{experiment}_targets.csv | |
recipe = | |
%{corpora} bnc dictionary \ | |
--corpus %{bnc_corpus} \ | |
-o %{target} \ | |
--stem \ | |
-v | |
# Count the word frequencies, without POS tags | |
[out/%{experiment}_dictionary-bnc-nopos.h5] | |
dep.targets = out/%{experiment}_targets.csv | |
recipe = | |
%{corpora} bnc dictionary \ | |
--corpus %{bnc_corpus} \ | |
-o %{target} \ | |
--stem \ | |
--omit-tags \ | |
-v | |
# Select only certain words, as the target tagged words | |
[out/%{experiment}_contexts_%{params}-pos_c-all-%{c_start}-%{c_end}.csv] | |
dep.file = out/%{experiment}_dictionary-%{params}-pos.h5 | |
recipe = | |
%{fowler_corpora_py} -c "import pandas as pd; pd.read_hdf('%{file}', key='dictionary')[%{c_start}:%{c_end}][['ngram', 'tag']].to_csv('%{target}', index=False)" | |
# Select only certain words, as the target untagged words (without POS) | |
[out/%{experiment}_contexts_%{params}-nopos_c-all-%{c_start}-%{c_end}.csv] | |
dep.file = out/%{experiment}_dictionary-%{params}-nopos.h5 | |
recipe = | |
%{fowler_corpora_py} -c "import pandas as pd; pd.read_hdf('%{file}', key='dictionary')[['ngram']].drop_duplicates()[%{c_start}:%{c_end}].to_csv('%{target}', index=False)" | |
# Build the space | |
[out/%{experiment}_space_%{params}.h5] | |
dep.targets = out/%{experiment}_targets.csv | |
dep.context = out/%{experiment}_contexts_%{params}.csv | |
recipe = | |
%{corpora} bnc cooccurrence \ | |
-t %{targets} \ | |
-c %{context} \ | |
--corpus %{bnc_corpus} \ | |
-o %{target} \ | |
--stem | |
# Run an experiment | |
[experiment_%{experiment}__%{params}] | |
type = task | |
dep.data = data/%{experiment}.csv | |
dep.space = out/%{experiment}_space_%{params}.h5 | |
recipe = | |
%{corpora} similarity %{experiment} \ | |
-s %{space} \ | |
--%{experiment}-data %{data} | |
[universe] | |
type = task | |
dep.a = experiment_wordsim353__bnc-pos_c-all-0-3101 | |
dep.b = experiment_wordsim353__bnc-pos_c-all-101-3101 | |
dep.c = experiment_wordsim353__bnc-nopos_c-all-0-3101 | |
dep.d = experiment_wordsim353__bnc-nopos_c-all-101-3101 | |
dep.e = experiment_rg65__bnc-pos_c-all-101-3101 | |
dep.f = experiment_rg65__bnc-nopos_c-all-101-3101 | |
[vacuum] | |
type = task | |
recipe = | |
rm -rf out/ data/ |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Hi
Can you please tell me how to execute this code?