Azure / azure-sdk-for-net

This repository is for active development of the Azure SDK for .NET. For consumers of the SDK we recommend visiting our public developer docs at https://learn.microsoft.com/dotnet/azure/ or our versioned developer docs at https://azure.github.io/azure-sdk-for-net.
MIT License
5.31k stars 4.66k forks source link

Consider naming for NamedEntity, LinkedEntity, and KeyPhrases #8894

Closed annelo-msft closed 4 years ago

annelo-msft commented 4 years ago

The names NamedEntity, LinkedEntity, and KeyPhrases relate more to the algorithms that identify them than what they actually represent.

Ideally, these would be more descriptive names, that indicate to users when to use which.

Key phrases are just the strings deemed important in the input text. It can be a noun or a verb, unlike entities in other cases. Given the algorithm, we may not be able to recognize the type. If we don't have key phrases, some entities ("important phrases") could be missed -- i.e. we'd overlook things that aren't in the type system used to train the NER model. People should use key phrases over the others when they want the full set of important words and they don't care about the type of the extracted phrase.

Named Entities are extracted based on the type system the NER model is trained with. This is currently people, organizations, and locations.

Linked Entities are used for disambiguation. They link to a knowledge base such as Wikipedia, and can tell you e.g. whether "bank" in "I'm going to the bank" is a river bank or a financial institution.

mayurid commented 4 years ago

Waiting for UX study

maririos commented 4 years ago

NamedEntity changes to => CategorizedEntity

maririos commented 4 years ago

Documentation should include literature naming

maririos commented 4 years ago

Key phrases should include an i.e. in the documentation so people know when to use it

maririos commented 4 years ago

For Linked Entities look #8781

maririos commented 4 years ago

For KeyPhrases see #9691