Communication routing based on user characteristics and behavior
An enhanced routing system determines a service provider best suited to fulfill a user's request to interact and establishes a communication session between the user's client device and a device of the service provider. The enhanced routing system may use user characteristics and behavior to select the service provider. For example, the enhanced routing system receives a request to connect to a customer service system from a user who has recently started a new job and has been accessing a banking application on his mobile phone. The enhanced routing system may determine that a payroll service provider is best suited to fulfill the user's request. For example, the enhanced routing system uses a machine learning model that has been trained on previously fulfilled requests. In this way, the enhanced routing system improves upon systems that continuously prompt the user for information by selecting a service provider without overburdening the user.
1. A method comprising:
receiving, from a client device of a user, a request to interact with a communications system;
selecting an entity within the communications system by applying a machine-learned model configured to a feature vector generated based on historic activity of the user, the machine-learned model configured to identify an entity of a plurality of entities of the communications system based at least in part on both a most recently viewed page associated with the communications system and a job title of the user; and
establishing a communication session between at least the client device and a device of the selected entity to fulfill the request to interact.
2. The method of claim 1 , further comprising receiving an identification number associated with the client device.
3. The method of claim 2 , wherein the identification number associated with the client device is at least one of a phone number, media access control (MAC) address, or internet protocol (IP) address.
4. The method of claim 2 , further comprising authenticating the client device comprises in response to determining that the identification number of the client device is a pre-approved number.
5. The method of claim 1 , wherein the historic activity of the user is associated with a user profile associated with the user, wherein the user profile is identified by:
accessing a personnel database comprising a plurality of user profiles; and
matching a phone number associated with the client device to a phone number associated with the user profile of the plurality of user profiles.
6. The method of claim 5 , wherein the user profile comprises at least one of a location of the user, information describing a relationship between the user and the communication system, the job title of the user, or satisfaction information authored by the user.
7. The method of claim 1 , wherein the historic activity of the user is determined by:
receiving activity history of the client device occurring within a time interval immediately preceding the request to interact with the communications system.
8. The method of claim 1 , wherein the historic activity of the user comprises at least one of applications executed by the client device, user actions occurring within an application executed by the client device, application or web pages viewed by the client device, and text entered into an input interface on the client device.
9. The method of claim 1 , wherein the request to interact with the communications system comprises a voice call from the client device.
10. The method of claim 1 , wherein the machine learning model is trained on data representative of previous interactions with the communications system by other users and user profiles and activity histories associated with the other users.
11. The method of claim 1 , wherein the entity is selected based at least in part on a dot product between the feature vector and a vector representative of characteristics of the entity.
12. The method of claim 1 , further comprising receiving a measure of satisfaction from the client device representative of the selected entity and re-training the machine learning model based at least in part on the received measure of satisfaction.
13. The method of claim 12 , wherein re-training the machine learning model comprises strengthening an association between the selected entity and characteristics of the user or historic activity of the user in response to the measure of satisfaction being a positive measure of satisfaction, and weakening the association between the selected entity and characteristics of the user or historic activity of the user in response to the measure of satisfaction being a negative measure of satisfaction.
14. The method of claim 12 , further comprising generating a data structure mapping the measure of satisfaction and information characterizing the communication session.
15. The method of claim 14 , wherein information characterizing the communication session comprises at least one of a list of participants in the communication session, a duration of the communication session, or a time at which the communication occurred.
16. The method of claim 14 , wherein re-training the machine learning model based at least in part on the received measure of satisfaction comprises re-training the machine learning model using the data structure.
17. The method of claim 13 , wherein the measure of satisfaction comprises a user selection of a different entity.
18. The method of claim 1 , wherein establishing the communication session between at least the client device and the device of the selected entity comprises initiating a phone or voice over IP call between the client device and the device of the selected entity.
19. A routing system configured to perform steps comprising:
receiving, from a client device of a user, a request to interact with a communications system;
selecting an entity within the communications system by applying a machine-learned model configured to a feature vector generated based on historic activity of the user, the machine-learned model configured to identify an entity of a plurality of entities of the communications system based at least in part on both a most recently viewed page associated with the communications system and a job title of the user; and
establishing a communication session between at least the client device and a device of the selected entity to fulfill the request to interact.
20. A non-transitory computer readable storage medium storing executable instructions that, when executed by one or more processors, cause the one or more processors to perform steps comprising:
receiving, from a client device of a user, a request to interact with a communications system;
selecting an entity within the communications system by applying a machine-learned model configured to a feature vector generated based on historic activity of the user, the machine-learned model configured to identify an entity of a plurality of entities of the communications system based at least in part on both a most recently viewed page associated with the communications system and a job title of the user; and
establishing a communication session between at least the client device and a device of the selected entity to fulfill the request to interact.