HW3 init

HW3/evaluation/README.md

@@ -0,0 +1,62 @@
+# ADL21-HW3
+Dataset & evaluation script for ADL 2021 homework 3
+
+## Dataset
+[download link](https://drive.google.com/file/d/186ejZVADY16RBfVjzcMcz9bal9L3inXC/view?usp=sharing)
+
+## Installation
+```
+git clone https://github.com/ntu-adl-ta/ADL21-HW3.git
+cd ADL21-HW3
+pip install -e tw_rouge
+```
+
+
+## Usage
+### Use the Script
+```
+usage: eval.py [-h] [-r REFERENCE] [-s SUBMISSION]
+
+optional arguments:
+  -h, --help            show this help message and exit
+  -r REFERENCE, --reference REFERENCE
+  -s SUBMISSION, --submission SUBMISSION
+```
+
+Example:
+```
+python eval.py -r public.jsonl -s submission.jsonl
+{
+  "rouge-1": {
+    "f": 0.21999419163162043,
+    "p": 0.2446195813913345,
+    "r": 0.2137398792982201
+  },
+  "rouge-2": {
+    "f": 0.0847583291303246,
+    "p": 0.09419044877345074,
+    "r": 0.08287844474014894
+  },
+  "rouge-l": {
+    "f": 0.21017939117006337,
+    "p": 0.25157090570020846,
+    "r": 0.19404349000921203
+  }
+}
+```
+
+
+### Use Python Library
+```
+>>> from tw_rouge import get_rouge
+>>> get_rouge('我是人', '我是一個人')
+{'rouge-1': {'f': 0.7499999953125, 'p': 1.0, 'r': 0.6}, 'rouge-2': {'f': 0.33333332888888895, 'p': 0.5, 'r': 0.25}, 'rouge-l': {'f': 0.7499999953125, 'p': 1.0, 'r': 0.6}}
+>>> get_rouge(['我是人'], [ 我是一個人'])
+{'rouge-1': {'f': 0.7499999953125, 'p': 1.0, 'r': 0.6}, 'rouge-2': {'f': 0.33333332888888895, 'p': 0.5, 'r': 0.25}, 'rouge-l': {'f': 0.7499999953125, 'p': 1.0, 'r': 0.6}}
+>>> get_rouge(['我是人'], ['我是一個人'], avg=False)
+[{'rouge-1': {'f': 0.7499999953125, 'p': 1.0, 'r': 0.6}, 'rouge-2': {'f': 0.33333332888888895, 'p': 0.5, 'r': 0.25}, 'rouge-l': {'f': 0.7499999953125, 'p': 1.0, 'r': 0.6}}]
+```
+
+
+## Reference
+[cccntu/tw_rouge](https://github.com/cccntu/tw_rouge)

HW3/evaluation/eval.py

@@ -0,0 +1,31 @@
+import json
+import argparse
+from tw_rouge import get_rouge
+
+
+def main(args):
+    refs, preds = {}, {}
+
+    with open(args.reference) as file:
+        for line in file:
+            line = json.loads(line)
+            refs[line['id']] = line['title']
+
+    with open(args.submission) as file:
+        for line in file:
+            line = json.loads(line)
+            preds[line['id']] = line['title']
+
+    keys =  refs.keys()
+    refs = [refs[key] for key in keys]
+    preds = [preds[key] for key in keys]
+
+    print(json.dumps(get_rouge(preds, refs), indent=2))
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    parser.add_argument('-r', '--reference')
+    parser.add_argument('-s', '--submission')
+    args = parser.parse_args()
+    main(args)

HW3/evaluation/tw_rouge/LICENSE

@@ -0,0 +1,10 @@
+
+The MIT License (MIT)
+Copyright (c) 2021, Jonathan Chang
+
+Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+

HW3/evaluation/tw_rouge/README.md

@@ -0,0 +1,17 @@
+# ROUGE score calculator with tradition chinese word segmentation
+
+## Install
+
+`pip install` this repo
+
+## Usage
+
+```
+>>> from tw_rouge import get_rouge
+>>> get_rouge('我是人', '我是一個人')
+{'rouge-1': {'f': 0.7499999953125, 'p': 1.0, 'r': 0.6}, 'rouge-2': {'f': 0.33333332888888895, 'p': 0.5, 'r': 0.25}, 'rouge-l': {'f': 0.7499999953125, 'p': 1.0, 'r': 0.6}}
+>>> get_rouge(['我是人'], [ 我是一個人'])
+{'rouge-1': {'f': 0.7499999953125, 'p': 1.0, 'r': 0.6}, 'rouge-2': {'f': 0.33333332888888895, 'p': 0.5, 'r': 0.25}, 'rouge-l': {'f': 0.7499999953125, 'p': 1.0, 'r': 0.6}}
+>>> get_rouge(['我是人'], ['我是一個人'], avg=False)
+[{'rouge-1': {'f': 0.7499999953125, 'p': 1.0, 'r': 0.6}, 'rouge-2': {'f': 0.33333332888888895, 'p': 0.5, 'r': 0.25}, 'rouge-l': {'f': 0.7499999953125, 'p': 1.0, 'r': 0.6}}]
+```

HW3/evaluation/tw_rouge/setup.py

@@ -0,0 +1,11 @@
+from setuptools import find_packages, setup
+
+setup(
+    name="tw_rouge",
+    packages=find_packages(),
+    version="0.1.0",
+    description="rouge metric for traditional chinese",
+    author="Jonathan Chang",
+    license="MIT",
+    install_requires=["ckiptagger[tf,gdown]", "rouge"],
+)

HW3/evaluation/tw_rouge/tw_rouge/__init__.py

@@ -0,0 +1 @@
+from .twrouge import get_rouge

HW3/evaluation/tw_rouge/tw_rouge/twrouge.py

@@ -0,0 +1,36 @@
+import os
+
+from ckiptagger import WS, data_utils
+from rouge import Rouge
+
+cache_dir = os.environ.get("XDG_CACHE_HOME", os.path.join(os.getenv("HOME"), ".cache"))
+download_dir = os.path.join(cache_dir, "ckiptagger")
+data_dir = os.path.join(cache_dir, "ckiptagger/data")
+os.makedirs(download_dir, exist_ok=True)
+if not os.path.exists(os.path.join(data_dir, "model_ws")):
+    data_utils.download_data_gdown(download_dir)
+
+ws = WS(data_dir)
+
+
+def tokenize_and_join(sentences):
+    return [" ".join(toks) for toks in ws(sentences)]
+
+
+rouge = Rouge()
+
+
+def get_rouge(preds, refs, avg=True, ignore_empty=False):
+    """wrapper around: from rouge import Rouge
+    Args:
+        preds: string or list of strings
+        refs: string or list of strings
+        avg: bool, return the average metrics if set to True
+        ignore_empty: bool, ignore empty pairs if set to True
+    """
+    if not isinstance(preds, list):
+        preds = [preds]
+    if not isinstance(refs, list):
+        refs = [refs]
+    preds, refs = tokenize_and_join(preds), tokenize_and_join(refs)
+    return rouge.get_scores(preds, refs, avg=avg, ignore_empty=ignore_empty)