def get_conn(self) -> LanguageServiceClient:
"""
Retrieves connection to Cloud Natural Language service.
:return: Cloud Natural Language service object
:rtype: google.cloud.language_v1.LanguageServiceClient
"""
if not self._conn:
self._conn = LanguageServiceClient(credentials=self._get_credentials(), client_info=CLIENT_INFO)
return self._conn
def _get_places(language_client: language_v1.LanguageServiceClient,
query: str) -> Sequence[language_v1.Entity]:
"""Returns a list of entities that are of type LOCATION."""
document = language_v1.Document(content=query,
type_=language_v1.Document.Type.PLAIN_TEXT)
response = language_client.analyze_entities(
request={
"document": document,
"encoding_type": language_v1.EncodingType.UTF8
})
locations = [
e for e in response.entities
if e.type == language_v1.Entity.Type.LOCATION and "mid" in e.metadata
]
return locations
def __init__(self):
self._gcp_client = LanguageServiceClient()
class EntitySentimentAnalyzer:
_logger = logging.getLogger("EntitySentimentAnalyzer")
def __init__(self):
self._gcp_client = LanguageServiceClient()
def perform_entity_sentiment_analysis(
self, docs: List[FOMCDoc]
) -> List[Tuple[FOMCDoc, Entity]]:
self._logger.info(
"Performing entity sentiment analysis on {} FOMC documents ...".format(
len(docs)
)
)
doc_to_sentiment_list = []
for fomc_doc in tqdm(docs):
document = Document(
content=fomc_doc.get_content(), type_=Document.Type.PLAIN_TEXT
)
response = self._gcp_client.analyze_entity_sentiment(document=document)
doc_to_sentiment_list.append((fomc_doc, response.entities))
return doc_to_sentiment_list
@classmethod
def get_entity_sentiment_overtime(
cls, entity_name: str, entity_sentiment_result: List[Tuple[FOMCDoc, Entity]]
) -> Union[pd.DataFrame, None]:
date_to_entity_sentiment = []
# TODO: aggregate lower and upper case entity names, aggregate by buckets
for fomc_doc, entities in entity_sentiment_result:
matching_entities = [
entity
for entity in entities
if entity_name.lower() == entity.name.lower() and entity.sentiment.score
]
if matching_entities:
if len(matching_entities) > 1:
cls._logger.info(
"Found {} entity {} times in the same doc".format(
entity_name, len(matching_entities)
)
)
date_to_entity_sentiment.append(
(fomc_doc.meeting_date, matching_entities[0].sentiment.score)
)
date_to_entity_sentiment = sorted(
date_to_entity_sentiment, key=lambda item: item[0]
)
if date_to_entity_sentiment:
dataframe = pd.DataFrame.from_records(date_to_entity_sentiment)
# dataframe.to_excel(os.path.join(OUTPUT_DIR, 'sentiment-for-{}-entity-overtime.xlsx'.format(entity_name)))
dataframe.set_index(dataframe[0], drop=True, inplace=True)
dataframe.drop(columns=[0], inplace=True)
dataframe.rename(columns={1: entity_name}, inplace=True)
return dataframe
else:
return None
@staticmethod
def pretty_print_entity(entity: Entity):
print(
"Name = {}, type = {}, magnitude = {}, score = {}, salience = {}".format(
entity.name,
entity.type_,
entity.sentiment.magnitude,
entity.sentiment.score,
entity.salience,
)
)