Skip to content

AutoML labeling tool is a UI to speed labeling text for AutoML NL applications. Using AutoML NL API syntax analysis, it applies annotations on a sentence by sentence basis. This tool speeds the human factor for annotation by using a simple interface. It runs independently as a client to the AutoML Natural Language API.

Notifications You must be signed in to change notification settings

byronwhitlock-google/AutoML-Labeling-Tool

Repository files navigation

AutoML-Labeling-Tool

ML labeling tool is a UI to assists labeling and relabeling text for entity extraction in Google AutoML Natural Language API. It speeds labeling of entities on a sentence by sentence basis. It runs independently as a client to the AutoML Natural Language API.

Vision

ML labeling tool is a UI to assist labeling text for entity extraction in Google AutoML Natural Language API. It speeds labeling of entities on a sentence by sentence basis. It runs independently as a client to the AutoML Natural Language API.

Motivation

Google Cloud AutoML Entity extraction can be trained to identify custom entities from a text string. For example, to train a model that takes the string "Jim bought 300 shares of Acme Corp. in 2006." as input and identifies the following entities: "Jim" as "Person", "Acme Corp." as "Organization", and "2006" as "Time" With longer document length texts, the entities should be a whole sentence, not a single word or short phrases.

The PSO ML labeling tool makes it to label or annotate a whole document and show the current prediction on a selected sentence. The prediction is validated or rejected by the SME one sentence at a time. Each predicted sentence contains multiple entities. The sentence accuracy is based on average accuracy of its contained entities. This allows a whole sentence to receive a label based on the extracted entities of the sentence.

UI Sample

User Interface Samle

About

AutoML labeling tool is a UI to speed labeling text for AutoML NL applications. Using AutoML NL API syntax analysis, it applies annotations on a sentence by sentence basis. This tool speeds the human factor for annotation by using a simple interface. It runs independently as a client to the AutoML Natural Language API.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 3

  •  
  •  
  •  

Languages