What's the training corpus of models behind GCP Natural Language APIs?

[aslidemiroz]

Hi, where can I find some information about which datasets are used for training models that power the natural language APIs for sentiment analysis, entity extraction, etc.? Thanks!

[comaro]

Natural Language API is trained using different types of datasets.

• Public datasets Examples: Five crowd-flower sentiment benchmarks
• EAP customer datasets Examples: Feefo sentiment dataset
• Academic datasets Examples: Stanford rotten tomatoes sentences, UCI Sentiment Labeled Sentences Data Set. See.
• Google datasets Examples: Shopping, Play.