Remote and local predictions

A method performed by a computing device, the method comprising:

• receiving, by the computing device, text input;
• sending, over a communications network, data indicative of the text input to a remote prediction engine having been trained to predict items based on text data;
• determining a number of n-grams in the text input;
• determining whether the number of n-grams exceeds a remote prediction threshold, the remote prediction threshold indicating a maximum number of n-grams for local processing by the computing device using a local prediction engine;
• in response to determining that the number of n-grams exceeds the remote prediction threshold, providing the received text input to the remote prediction engine and providing, by the computing device, data indicative of the text to the local prediction engine executing at the computing device;
• monitoring, by the computing device, based on determining that the number of n-grams exceeds the remote prediction threshold, for a remote predicted item from the remote prediction engine;
• in response to receiving the remote predicted item within a time interval determined from the time at which the text input is received, generating a final predicted item based on the remote predicted item or in response to not receiving the remote predicted item within the time interval, generating the final predicted item based on a local predicted item from the local prediction engine;
• and
• presenting the final predicted item on the computing device.

At least one non-transitory computer readable medium including instructions that when executed by at least one processor of a computing device, cause the at least one processor to:

• receive text input;
• send, over a communications network, data indicative of the text input to a remote prediction engine having been trained to predict items based on text;
• determine a number of n-grams in the text input;
• determine whether the number of n-grams exceeds a remote prediction threshold, the remote prediction threshold indicating a maximum number of n-grams for local processing by the computing device using a local prediction engine;
• in response to determining that the number of n-grams exceeds the remote prediction threshold, provide the received text input to the remote prediction engine and provide the received text input to the local prediction engine executing at the computing device;
• monitor, based on the determination that the number of n-grams exceeds the remote prediction threshold, for a remote predicted item from the remote prediction engine;
• wherein the computing device is configured to, in response to receiving the remote predicted item within a time interval determined from the time at which the text input is received, generate a final predicted item based on the remote predicted item and in response to not receiving the remote predicted item within the time interval, generate the final predicted item based on a local predicted item from the local prediction engine;
• and
• present, on a user interface of the computing device, the final predicted item, the user interface operable to receive an acknowledgment to enter the final predicted item into the computing device.