Determining device assistant manner of reply

A computer program product for replying to user commands, the computer program product comprising:

• one or more non-transitory computer-readable storage media and program instructions stored on the one or more non-transitory computer-readable storage media capable of performing a method, the method comprising:
• receiving a command for information from a first user of one or more devices, wherein the one or more devices are associated with at least two users;
• collecting preferences of the first user from a database and collecting user context from the one or more devices;
• extracting a plurality of features of the command based on the preferences and the user context collected, wherein the plurality of features include at least a topic and an activity corresponding to the command;
• identifying an appropriate manner of reply for the command using one or more machine learning models based on the plurality of features extracted, wherein the one or more machine learning models is trained to weight the plurality of features according to a correlation with the appropriate manner of reply, and wherein the appropriate manner of reply indicates a completeness of the information to include in an appropriate reply to the first user;
• generating the appropriate reply to the command based on an identified manner of reply for the command using the one or more machine learning models, wherein the appropriate reply to the command is a hint, wherein prior to a transmitting of the appropriate reply to the first user of the one or more devices, notifying an administrator of the appropriate reply, and receiving a confirmation or approval of the appropriate reply from the administrator;
• transmitting the appropriate reply to the first user of the one or more devices following the approval of the appropriate reply by the administrator;
• determining a usefulness of the hint based on feedback received from the first user, wherein the feedback is utilized in adjusting the one or more machine learning models for the first user;
• anticipating a new command for information from a second user based on the second user entering a geofence of the one or more devices;
• collecting preferences of the second user from the database and user context data for the second user from the one or more devices;
• retrieving, from a queue, at least two or more hints and an answer to the new command based on the plurality of appropriate replies identified for the new command using the one or more machine learning models, wherein the one or more machine learning models utilize at least the preferences of the second user and the user context data for the second user collected from the one or more devices in anticipation of the new command;
• transmitting a first hint of the two or more hints to the second user;
• transmitting, in response to negative feedback received for the first hint, a second hint of the two or more hints to the second user;
• transmitting the answer to the second user of the one or more devices;
• and
• determining whether the answer to the new command is helpful to the second user based on an interaction of the second user with the answer, retraining the one or more machine learning models for the second user based on the interaction of the second user with the answer and the feedback associated with the at least two or more hints.

A computer system for replying to user commands, the computer system comprising:

• one or more computer processors, one or more computer-readable storage media, and program instructions stored on the one or more of the computer-readable storage media for execution by at least one of the one or more processors capable of performing a method, the method comprising:
• receiving a command for information from a first user of one or more devices, wherein the one or more devices are associated with at least two users;
• collecting preferences of the first user from a database and collecting user context from the one or more devices;
• extracting a plurality of features of the command based on the preferences and the user context collected, wherein the plurality of features include at least a topic and an activity corresponding to the command;
• identifying an appropriate manner of reply for the command using one or more machine learning models based on the plurality of features extracted, wherein the one or more machine learning models is trained to weight the plurality of features according to a correlation with the appropriate manner of reply, and wherein the appropriate manner of reply indicates a completeness of the information to include in an appropriate reply to the first user;
• generating the appropriate reply to the command based on an identified manner of reply for the command using the one or more machine learning models, wherein the appropriate reply to the command is a hint, wherein prior to a transmitting of the appropriate reply to the first user of the one or more devices, notifying an administrator of the appropriate reply, and receiving a confirmation or approval of the appropriate reply from the administrator;
• transmitting the appropriate reply to the first user of the one or more devices following the approval of the appropriate reply by the administrator;
• determining a usefulness of the hint based on feedback received from the first user, wherein the feedback is utilized in adjusting the one or more machine learning models for the first user;
• anticipating a new command for information from a second user based on the second user entering a geofence of the one or more devices;
• collecting preferences of the second user from the database and user context data for the second user from the one or more devices;
• retrieving, from a queue, at least two or more hints and an answer to the new command based on the plurality of appropriate replies identified for the new command using the one or more machine learning models, wherein the one or more machine learning models utilize at least the preferences of the second user and the user context data for the second user collected from the one or more devices in anticipation of the new command;
• transmitting a first hint of the two or more hints to the second user;
• transmitting, in response to negative feedback received for the first hint, a second hint of the two or more hints to the second user;
• transmitting the answer to the second user of the one or more devices;
• and
• determining whether the answer to the new command is helpful to the second user based on an interaction of the second user with the answer, retraining the one or more machine learning models for the second user based on the interaction of the second user with the answer and the feedback associated with the at least two or more hints.