def main() -> list:
current_dir = Path(path.dirname(path.realpath(__file__)))
inference = SentimentInference(current_dir / 'aspects.csv',
current_dir / 'opinions.csv')
print('\n' + '=' * 40 + '\n' +
'Running inference on examples from sample test set:\n')
docs = [
"The food was very fresh and flavoursome the service was very attentive. Would go back"
" to this restaurant if visiting London again.",
"The food was wonderful and fresh, I really enjoyed it and will definitely go back. "
"Staff were friendly.",
"The ambiance is charming. Uncharacteristically, the service was DREADFUL. When we"
" wanted to pay our bill at the end of the evening, our waitress was nowhere to be "
"found..."
]
sentiment_docs = []
for doc_raw in docs:
print('Raw Document: \n{}'.format(doc_raw))
sentiment_doc = inference.run(doc=doc_raw)
sentiment_docs.append(sentiment_doc)
print('SentimentDocument: \n{}\n'.format(sentiment_doc) + '=' * 40 +
'\n')
return sentiment_docs
def test_inference():
lexicons_dir = Path(LIBRARY_ROOT) / 'examples' / 'absa'
inference = SentimentInference(lexicons_dir / 'aspects.csv', lexicons_dir / 'opinions.csv',
parse=False)
data_dir = Path(LIBRARY_ROOT) / 'tests' / 'fixtures' / 'data' / 'absa'
for i in range(1, 4):
with open(data_dir / 'core_nlp_doc_{}.json'.format(i)) as f:
predicted_doc = inference.run(parsed_doc=json.load(f, object_hook=CoreNLPDoc.decoder))
with open(data_dir / 'sentiment_doc_{}.json'.format(i)) as f:
expected_doc = json.load(f, object_hook=SentimentDoc.decoder)
assert expected_doc == predicted_doc
def run(
self,
aspect_lex: PathLike = None,
opinion_lex: PathLike = None,
data: PathLike = None,
parsed_data: PathLike = None,
inference_results: PathLike = None,
) -> Optional[pd.DataFrame]:
opinions = load_opinion_lex(opinion_lex)
if not opinions:
raise ValueError("Empty opinion lexicon!")
aspects = pd.read_csv(aspect_lex, header=None, encoding="utf-8")[0]
if aspects.empty:
raise ValueError("Empty aspect lexicon!")
if inference_results:
with open(inference_results, encoding="utf-8") as f:
results = json.loads(f.read(),
object_hook=SentimentDoc.decoder)
elif data or parsed_data:
inference = SentimentInference(aspect_lex, opinions, parse=False)
parse = None
if not parsed_data: # source data is raw text, need to parse
from nlp_architect.pipelines.spacy_bist import SpacyBISTParser
parse = SpacyBISTParser().parse
results = {}
print("Running inference on data files... (Iterating data files)")
data_source = parsed_data if parsed_data else data
for file, doc in self._iterate_docs(data_source):
parsed_doc = (parse(doc) if parse else json.loads(
doc, object_hook=CoreNLPDoc.decoder))
sentiment_doc = inference.run(parsed_doc=parsed_doc)
if sentiment_doc:
results[file] = sentiment_doc
with open(SENTIMENT_OUT / "inference_results.json",
"w",
encoding="utf-8") as f:
json.dump(results,
f,
cls=SentimentDocEncoder,
indent=4,
sort_keys=True)
else:
print(
"No input given. Please supply one of: "
"data directory, parsed data directory, or inference results.")
return None
print("\nComputing statistics...")
stats = self._compute_stats(results, aspects, opinions)
print("Done.")
return stats
def main() -> None:
parser = argparse.ArgumentParser(description='ABSA Inference')
parser.add_argument('--aspects', type=validate_existing_path,
help='Path to aspect lexicon (csv)', required=True)
parser.add_argument('--opinions', type=validate_existing_path, required=True,
help='Path to opinion lexicon (csv)')
args = parser.parse_args()
inference = SentimentInference(aspect_lex=args.aspects, opinion_lex=args.opinions)
while True:
doc = input('\nEnter sentence >> ')
print(inference.run(doc))
def test_inference():
lexicons_dir = Path(LIBRARY_ROOT) / "examples" / "absa"
inference = SentimentInference(lexicons_dir / "aspects.csv",
lexicons_dir / "opinions.csv",
parse=False)
data_dir = Path(LIBRARY_ROOT) / "tests" / "fixtures" / "data" / "absa"
for i in range(1, 4):
with open(data_dir / "core_nlp_doc_{}.json".format(i)) as f:
predicted_doc = inference.run(
parsed_doc=json.load(f, object_hook=CoreNLPDoc.decoder))
with open(data_dir / "sentiment_doc_{}.json".format(i)) as f:
expected_doc = json.load(f, object_hook=SentimentDoc.decoder)
assert expected_doc == predicted_doc
def main() -> list:
inference = SentimentInference("/home/coeuser/Desktop/absa/examples/generated_aspect_lex_updated_v3.csv", "/home/coeuser/Desktop/absa/examples/generated_opinion_lex_reranked_v3.csv", parse = True)
sentiment_docs = []
with open('/home/coeuser/Desktop/absa/examples/step1_competitor_news_data.txt', 'r', encoding = 'latin-1') as csv:
lines = csv.readlines()
for id, line in enumerate(tqdm(lines)):
if line:
line = line.split('\t')
sentiment_doc = inference.run(doc = line[2])
if sentiment_doc != None:
labels = doc2label(sentiment_doc)
labels['sent_id'] = line[0]
labels['_vendor_name'] = line[1]
labels = labels_enhancer(labels)
with open('/home/coeuser/Desktop/absa/examples/' + 'step2_competitor_news_data_sentiment_scores.json', 'a') as json_file:
json_file.write(json.dumps(labels))
json_file.write('\n')
inference = SentimentInference(
LEXICONS_OUT / 'generated_aspect_lex.csv',
LEXICONS_OUT / 'generated_opinion_lex_reranked.csv')
clothing_val = os.path.join(args.data_folder,
'clothing_data/clothing-absa-validation.json')
with open(clothing_val) as json_file:
val = json.load(json_file)
predictions = []
vals = []
for doc in val["data"]:
doc_raw = " ".join([token[0] for token in doc])
sentiment_doc = inference.run(doc=doc_raw)
if sentiment_doc is not None:
predictions.append(doc2IO(sentiment_doc))
vals.append(doc)
y_pred = flatten(predictions)[1::2]
y_true = flatten(vals)[1::2]
from sklearn.metrics import f1_score
# Log metrics
run = Run.get_context()
run.log('Aspect Lexicon Size', len(aspect_lex))
run.log('Opinion Lexicon Size', len(opinion_lex))
run.log('f1_weighted', float(f1_score(y_true, y_pred, average='weighted')))
documents['_sentences'] = sentences_list
return documents
inference = SentimentInference(aspect_path, opinion_path)
shutil.copyfile(aspect_path, './outputs/news_content_aspect.csv')
shutil.copyfile(opinion_path, './outputs/news_content_opinion.csv')
input_file_path = os.path.join(args.data_folder,
'news_data/all_news_content.csv')
print(
f'Aspect and Opinion lexicons files loaded from {aspect_path} and {opinion_path}'
)
print(f'Input file loaded from {input_file_path}')
inference = SentimentInference(aspect_path, opinion_path)
# Get Inference Results
with open(input_file_path, 'r') as csv_file:
lines = csv_file.readlines()
for id, line in enumerate(lines):
if line:
sentiment_doc = inference.run(doc=line)
if sentiment_doc != None:
labels = doc2label(sentiment_doc)
labels['sent_id'] = id + 1
with open('./outputs/' + 'sentiment_labels_v1.json',
'a') as json_file:
json_file.write(json.dumps(labels))
json_file.write('\n')