Providing and leveraging implicit signals reflecting user-to-BOT interaction

A method for assisting users in interacting with BOTs, comprising:

• receiving one or more current-transaction implicit quality signals produced in a course of interaction between a current user and a current BOT,
• the current BOT corresponding to a computer-implemented agent that provides a service to the current user, and
• each current-transaction implicit quality signal reflecting a quality of service provided by the current BOT to the current user;
• using an abandonment-determination component provided by hardware logic circuitry to map said one or more current-transaction implicit quality signals to an output result, the output result constituting a BOT-recommendation event, wherein the BOT-recommendation event indicates that the current user has abandoned a transaction with the current BOT, or is likely to abandon the transaction with the current BOT;
• in response to the BOT-recommendation event, identifying, using a recommendation engine provided by the hardware logic circuitry, at least one recommended BOT in a collection of BOTs as a replacement to the current BOT, to provide a recommendation result, said at least one recommended BOT that is identified being different from the current BOT;
• sending an electronic message to a user computing device associated with the current user, the electronic message conveying the recommendation result;
• and
• invoking a selected BOT in response to a BOT activation event.

A computer-readable storage medium for storing computer-readable instructions, the computer-readable instructions, when executed by one or more hardware processors, performing a method that comprises:

• providing implicit quality signals over a span of time that reflect quality of service provided by a collection of BOTs to a group of users in a course of performing transactions;
• storing the implicit quality signals in a signal data store,
• the BOTs corresponding to computer-implemented agents that provide respective services to the group of users,
• said providing involving capturing transaction information related to transaction-related events that occur over a course communication between the group of users and the
• the implicit quality signals including at least:
• one of more user-behavior implicit signals, each of which describes one or more actions taken by a given user in interacting with a given BOT, and each of which has an implicit bearing on quality of service provided by the given BOT;
• and
• one of more BOT-behavior implicit signals, each of which describes one or more actions taken by the given BOT in interacting with the given user, and each of which has an implicit bearing on quality of service provided by the given BOT;
• training a machine-trained model based, at least in part, on a training data set, the training data set including features associated with the transactions performed between the group of users and the collection of BOTs, together with evidence of whether these transactions culminated in abandonment of the transactions, the features including the implicit quality signals provided in the signal data store;
• and
• transferring the machine-trained model to an abandonment-determination component for use by the abandonment-determination component in determining whether a current user has abandoned a transaction with a current BOT, or is likely to abandon the transaction with the current BOT, the abandonment-determination component being implemented by the computer-readable instructions stored on the computer-readable storage medium,
• wherein the abandonment-determination component triggers a recommendation engine to identify at least one recommended BOT as a replacement to the current BOT, said at least one recommended BOT that is identified being different from the current BOT, the recommendation engine being implemented by the computer-readable instructions stored on the computer-readable storage medium.