Patent attributes
A method for dynamically updating a machine learning model is provided. The method comprises receiving first data comprising a first number of documents, training a first machine learning model with the first data, and releasing the first machine learning model for use. After receipt of the first data, second data is received comprising a second number of documents. A concept labeling engine, according to the first machine learning model, labels a concept for at least one document in the second data. Responsive to determining that a distribution of words in the second data deviates from a distribution of words in the first data beyond a specified threshold, the first machine learning model is retrained with at least part of the second data, the retraining producing a second machine learning model; and releasing the second machine learning model for use.
