Adobe Inc.
Self-trained content management system for automatically classifying execution modes for user requests

Abstract:

Embodiments of the present invention are directed to techniques for optimizing an execution mode used to process a request. A self-training storage system can determine one or more proposed execution modes based on a real-time evaluation of various factors, including a history of execution modes used to process historical requests. This history can serve as training data for a machine learning model that can predict an execution mode based on one or more request parameters. This predicted execution mode can be taken as one of a number of proposed execution modes that can be combined using a weightage system to automatically classify an ultimate execution mode used to process the request. Associated weights for combining proposed execution modes can be determined by training a machine learning model using a training set constructed based on a survey of user satisfaction with a determined execution mode.