fix bug

Author: eleme

data_preprocess/create_kg.py

@@ -48,7 +48,7 @@ def fetch_triple(fb_file, dataset_dir, out_dir):
                     triples[subject] = [(relation, www2fb(obj))]
                 if subject not in sub2rel:
                     sub2rel[subject] = set()
-                sub2rel.add(relation)
+                sub2rel[subject].add(relation)
 
     # agument with triples from dataset
     data_files = [os.path.join(os.path.dirname(dataset_dir), file) for file in os.listdir(dataset_dir)]

data_preprocess/fetch.sh

@@ -1,27 +1,32 @@
+#!/usr/bin/env bash
 # download dataset and put it in data directory
 
 cd ..
 mkdir data
 mkdir raw_data
 cd ./raw_data
 
-echo "Downloading SimpleQuestions dataset...\n"
-wget https://www.dropbox.com/s/tohrsllcfy7rch4/SimpleQuestions_v2.tgz
+printf "\nDownloading SimpleQuestions dataset...\n"
+#wget https://www.dropbox.com/s/tohrsllcfy7rch4/SimpleQuestions_v2.tgz
 
-echo "\n\nUnzipping SimpleQuestions dataset...\n"
+printf "\nUnzipping SimpleQuestions dataset...\n"
 tar -xvzf SimpleQuestions_v2.tgz
 rm SimpleQuestions_v2.tgz
 
 cd ./SimpleQuestions_v2
 mkdir dataset
-mv *.txt ./dataset
+mv annotated*.txt ./dataset
 
 cd ..
-echo "Downloading FB5M-extra...\n
-wget https://www.dropbox.com/s/dt4i1a1wayks43n/FB5M-extra.tar.gz
+printf "\nDownloading FB5M-extra...\n"
+#wget https://www.dropbox.com/s/dt4i1a1wayks43n/FB5M-extra.tar.gz
 
-echo "\n\nUnzipping FB5M-extra...\n"
+echo "\nUnzipping FB5M-extra...\n"
 tar -xzf FB5M-extra.tar.gz
 rm FB5M-extra.tar.gz
+mkdir FB5M-extra
+mv FB5M.en-name.txt FB5M-extra/
+mv FB5M.name.txt FB5M-extra/
+mv FB5M.type.txt FB5M-extra/
 
 

data_preprocess/preprocess.sh

@@ -1,3 +1,4 @@
+#!/usr/bin/env bash
 echo "\n\n create knowledge graph\n"
 python create_kg.py ../raw_data/SimpleQuestions_v2/freebase-subsets/freebase-FB5M.txt ../raw_data/SimpleQuestions_v2/dataset/ ../data
 

models/TransE/train.sh

@@ -1,3 +1,4 @@
+#!/usr/bin/env bash
 CUDA_VISIBLE_DEVICES=2 \
 python main.py \
 --data_dir ../../data \

models/inference/start.sh

@@ -1,3 +1,4 @@
+#!/usr/bin/env bash
 python inference.py \
 --entdect_type "bilstm_crf" \
 --subnet_type "transe" \