IssueTracker

Currently, NLP Entity Extraction product misclassifies the Amazon Rainforest as the Amazon company. The model should be able to distinguish between them as they are rather different.

Steps to reproduce:

-1: Using the python library, make use of language.LanguageServiceClient().analyze_entities() with encoding_type='UTF32'.
The document provided is set as type: language.types.Document(type='PLAIN_TEXT')

-2: Analyse the following script:

"One year on, family of murdered Amazon land defender say nothing has changed By Zoe Tabary, Thomson Reuters Foundation Nov 1 (Thomson
 Reuters Foundation) - Illegal loggers are operating unchecked in Brazil’s Amazon as coronavirus rages despite government vows to act after an indigenous land
 activist was killed exactly one year ago trying to protect the rainforest, said his cousin. Paulo Paulino Guajajara, or Lobo - ‘wolf’ in Portuguese - was hun
ting on Nov. 1, 2019, inside the Arariboia reservation in Maranhao state when he was attacked and shot in the head. His cousin, Laercio Guajajara, was wounded
 but escaped. Lobo, aged about 27, and Laercio were part of a group of 120 Guajajara Indian volunteers, brigades known as 'Guardians of the Forest', who patro
l protected areas of the world’s largest rainforest, burning the camps and timber of illegal loggers. 'The government said it would protect our land but nothi
ng has changed. The logging continues,' Laercio told the Thomson Reuters Foundation by phone, requesting his location not be disclosed for security reasons. “
One year after …"

-3: The results returned by the NLP are: Amazon, Amazon_company of type ORGANIZATION