edit tool and add result analysis

5 files changed, 71 insertions, 0 deletions

diff --git a/classification/tools/create_csv.py b/classification/tools/create_csv.py
index c57c45e6..ea997ccf 100755
--- a/classification/tools/create_csv.py
+++ b/classification/tools/create_csv.py
@@ -1,7 +1,13 @@
 from os import listdir, path
+from argparse import ArgumentParser
 
 root_directory = "../../results/classifier/"
 
+parser = ArgumentParser()
+parser.add_argument('-d', '--directory')
+
+args = parser.parse_args()
+
 def parse_iteration(directory):
     dictionary = {}
 
@@ -19,6 +25,11 @@ def output_csv(dictionary, full_path):
             file.write(f"{key}, {value}\n")
 
 def main():
+    if args.directory:
+        dictionary = parse_iteration(args.directory)
+        output_csv(dictionary, args.directory)
+        exit()
+
     for entry in listdir(root_directory):
         full_path = path.join(root_directory, entry)
         if path.isdir(full_path):
diff --git a/results/classifier/deepseek-2/analysis.csv b/results/classifier/deepseek-2/analysis.csv
new file mode 100644
index 00000000..bf541a72
--- /dev/null
+++ b/results/classifier/deepseek-2/analysis.csv
@@ -0,0 +1,9 @@
+category, count
+assembly, 5
+mistranslation, 47
+peripherals, 1
+manual-review, 5
+KVM, 1
+permissions, 1
+hypervisor, 1
+other, 1
diff --git a/results/classifier/deepseek-2/categories.csv b/results/classifier/deepseek-2/categories.csv
new file mode 100644
index 00000000..8c25cc62
--- /dev/null
+++ b/results/classifier/deepseek-2/categories.csv
@@ -0,0 +1,19 @@
+category, count
+vnc, 49
+performance, 102
+network, 370
+graphic, 388
+device, 881
+permissions, 38
+other, 708
+manual-review, 749
+boot, 189
+kernel, 54
+hypervisor, 654
+KVM, 146
+files, 226
+debug, 98
+assembly, 75
+mistranslation, 691
+peripherals, 274
+socket, 37
diff --git a/results/classifier/gemma3:12b/analysis.csv b/results/classifier/gemma3:12b/analysis.csv
new file mode 100644
index 00000000..4b6e8725
--- /dev/null
+++ b/results/classifier/gemma3:12b/analysis.csv
@@ -0,0 +1,13 @@
+category, count
+assembly, 19
+mistranslation, 21
+device, 4
+debug, 1
+peripherals, 1
+performance, 1
+kernel, 3
+hypervisor, 6
+network, 1
+kvm, 3
+permissions, 1
+boot, 1
diff --git a/results/classifier/gemma3:12b/categories.csv b/results/classifier/gemma3:12b/categories.csv
new file mode 100644
index 00000000..ebf2a416
--- /dev/null
+++ b/results/classifier/gemma3:12b/categories.csv
@@ -0,0 +1,19 @@
+category, count
+vnc, 74
+performance, 233
+network, 386
+graphic, 359
+device, 950
+permissions, 111
+other, 254
+kvm, 397
+manual-review, 115
+boot, 325
+kernel, 455
+hypervisor, 478
+files, 459
+debug, 321
+assembly, 152
+mistranslation, 136
+peripherals, 452
+socket, 72