Hybrid nlp scenarios for mobile devices
AssigneeSap SE
InventorsJan Gabriel, Sebastian Netuschil
Life of this patent
Estimating the term…
Abstract
A system and method include reception, from a remote system, of a trained natural language processing model implementing a dialog flow-based application, reception of input parameters from an operator while not in communication with the remote system, input of the input parameters to the model to initiate a transaction, storing of transaction data associated with the transaction in a local memory, determination that the mobile device is in communication with the remote system, transmission, in response to the determination, of the stored transaction data to the remote system, and reception, from the remote system, of an updated trained model based on the transmitted transaction data.
Claims20
An apparatus comprising:
- a communication device to receive, from a remote system, a trained model implementing a dialog flow-based application;
- an input device to receive input parameters from an operator;
- a local memory;
- and
- a processor to execute processor-executable process steps to cause the apparatus to:
- input the input parameters to the model to initiate a dialog while not in communication with the remote system;
- store data associated with the dialog in the local memory;
- determine that the mobile device is in communication with the remote system;
- in response to the determination, transmit the stored data to the remote system;
- and
- receive an updated trained model based on the transmitted data from the remote system via the communication device.
A method for a mobile device, comprising:
- receiving, from a remote system, a trained natural language processing model implementing a dialog flow-based application;
- receiving input parameters from an operator while not in communication with the remote system;
- inputting the input parameters to the model to initiate a dialog;
- storing data associated with the dialog in a local memory;
- determining that the mobile device is in communication with the remote system;
- in response to the determination, transmitting the stored data to the remote system;
- and
- receiving, from the remote system, an updated trained model based on the transmitted data.
A system comprising:
- an application platform to generate a first trained model of a natural language dialog flow and a second trained model of the natural language dialog flow, the first trained model for execution by a first operating system and the second trained model for execution by a second operating system;
- a first mobile device comprising a first local memory and executing the first operating system, the first mobile device to:
- receive the first trained model from the application platform;
- receive input parameters from a first operator;
- input the input parameters to the first trained model to initiate a first dialog;
- store first data associated with the first dialog in the first local memory;
- and
- transmit the stored first data to the application platform;
- and
- a second mobile device comprising a second local memory and executing the second operating system, the second mobile device to:
- receive the second trained model from the application platform;
- receive second input parameters from a second operator;
- input the second input parameters to the second trained model to initiate a second dialog;
- store second data associated with the second dialog in the second local memory;
- and
- transmit the second data to the application platform,
- wherein the application platform is to update the first trained model and the second trained model based on the first data and the second data.
Dependencies are inferred from the claim text.