Serialization fixed in build_deepwalk_corpus_iter() for non-str walk items

README.rst

@@ -4,6 +4,8 @@ DeepWalk
 
 DeepWalk uses short random walks to learn representations for vertices in graphs.
 
+  This implementation extends the original `DeepWalk <https://github.com/phanein/deepwalk>`_ v.1.0.3 with numerical walk items besides the ``str``, which is required by `HARP (AAAI 2018) <https://github.com/eXascaleInfolab/HARP>`_ and overall provides efficient graph node embedding, where the nodes are specified by numerical ids.
+
 Usage
 -----
 

deepwalk/walks.py

@@ -1,3 +1,4 @@
+from __future__ import print_function, division
 import logging
 from io import open
 from os import path
@@ -55,7 +56,7 @@ def _write_walks_to_disk(args):
   with open(f, 'w') as fout:
     for walk in graph.build_deepwalk_corpus_iter(G=G, num_paths=num_paths, path_length=path_length,
                                                  alpha=alpha, rand=rand):
-      fout.write(u"{}\n".format(u" ".join(v for v in walk)))
+      print(" ".join(str(v) for v in walk), file=fout)
   logger.debug("Generated new file {}, it took {} seconds".format(f, time() - t_0))
   return f
 
@@ -82,8 +83,13 @@ def write_walks_to_disk(G, filebase, num_paths, path_length, alpha=0, rand=rando
         files.append(file_)
 
   with ProcessPoolExecutor(max_workers=num_workers) as executor:
-    for file_ in executor.map(_write_walks_to_disk, args_list):
-      files.append(file_)
+    file_ = None
+    try:
+      for file_ in executor.map(_write_walks_to_disk, args_list):
+        files.append(file_)
+    except TypeError as err:
+      logger.error('ERROR: {}, file_: {}, args_list: {}'.format(err, file_, args_list))
+      raise
 
   return files
 

run.sh

@@ -0,0 +1,58 @@
+#!/bin/bash
+#
+# \description  Execution on multiple networks
+#
+# \author Artem V L <[email protected]>  https://exascale.info
+
+DIMS=128
+NETS="blogcatalog dblp homo wiki youtube"
+WORKERS=8
+RESTRACER=./exectime  # time
+LOGDIR=embeds/logs
+mkdir -p $LOGDIR
+
+USAGE="$0 -h | [-d <dimensions>=${DIMS}] [-w <workers>=${WORKERS}]
+  -d,--dims  - required number of dimensions in the embedding model
+  -w,--workers  - maximal number of workers (parallel thread). Note: deepwalk training can be failed on non-small datasets with small number of workers
+  -h,--help  - help, show this usage description
+
+  Examples:
+  \$ $0 -d 128 -w 4
+"
+
+while [ $1 ]; do
+	case $1 in
+	-h|--help)
+		# Use defaults for the remained parameters
+		echo -e $USAGE # -e to interpret '\n\
+		exit 0
+		;;
+	-d|--dims)
+		if [ "${2::1}" == "-" ]; then
+			echo "ERROR, invalid argument value of $1: $2"
+			exit 1
+		fi
+		DIMS=$2
+		echo "Set $1: $2"
+		shift 2
+		;;
+	-w|--workers)
+		if [ "${2::1}" == "-" ]; then
+			echo "ERROR, invalid argument value of $1: $2"
+			exit 1
+		fi
+		WORKERS=$2
+		echo "Set $1: $2"
+		shift 2
+		;;
+	*)
+		printf "Error: Invalid option specified: $1 $2 ...\n\n$USAGE"
+		exit 1
+		;;
+	esac
+done
+
+for NET in $NETS; do
+	$RESTRACER python3 -m deepwalk --format mat --input  graphs/${NET}.mat --number-walks 80 --representation-size ${DIMS} --walk-length 40 --window-size 10 --workers $WORKERS --output embeds/embs_${NET}_${DIMS}-n80-l40-s10.w2v > "$LOGDIR/${NET}_${DIMS}-n80-l40-s10.log" 2> "$LOGDIR/${NET}_${DIMS}-n80-l40-s10.err" # &
+done
+# python3 -m deepwalk --format mat --input example_graphs/blogcatalog.mat --number-walks 80 --representation-size 128 --walk-length 40 --window-size 10 --workers 1 --output example_graphs/blogcatalog.w2v