WRITING PERSONALIZED ELECTRONIC MESSAGES USING TEMPLATE-BASED AND MACHINE-LEARNING APPROACHES

• US 20200134009A1
• Filed: 10/31/2018
• Published: 04/30/2020
• Est. Priority Date: 10/31/2018
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Zhao, Bing, Rosales-Delmoral, Romer E., Li, Baolei
• Granted Patent: US 10,990,754 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 16/93 Document management systems
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06F 40/186 Templates
• CPC Class Codes: G06N 3/044 Recurrent networks, e.g. Hopfield networks
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: H04L 51/02 using automatic reactions or user delegation, e.g. automatic replies or chatbot-generated messages
• CPC Class Codes: H04L 51/52 for supporting social networking services

Abstract

This disclosure provides for systems and methods that generate personalized electronic messages for members of a networked communication service. The personalized electronic messages are generated according to commonalities between member profiles. In one approach, an electronic message template is referenced based on the commonalities. The electronic message template is populated with member profile attribute values selected from the member profiles. In another approach, a long-short term memory (LSTM) recurrent neural network (RNN) is used to generate the electronic messages. Under this approach, a sequence-to-sequence model is trained using previous electronic messages labeled with one or more member profile attributes and/or member profile attribute values. When provided with one or more member profile attribute values associated with matching member profiles, the LSTM RNN outputs a relevant and appropriate electronic message designed to create an interest in the recipient of the electronic message.

Claims

1. A system comprising:

a machine-readable medium storing computer-executable instructions; and

at least one hardware processor communicatively coupled to the machine-readable medium that, when the computer-executable instructions are executed, configures the system to;

select a first member profile of a networked communication service;

determine a plurality of member profiles from the networked communication service based on the selected first member profile;

receive a selection of a second member profile, the second member profile being selected from the first plurality of member profiles;

receive an instruction to generate an electronic message for sending to a member of the networked communication service associated with the selected second member profile;

match a first plurality of member profile attribute values associated with the first member profile with a second plurality of member profile attribute values associated with the second member profile to obtain a plurality of matched member profile attribute values;

generate a personalized electronic message for the member associated with the selected second member profile based on the plurality of matched member profile attribute values; and

send the personalized electronic message to the member associated with the selected second member profile.

2. The system of claim 1, wherein the system is configured to generate the personalized electronic message by:

identifying an electronic message template from a plurality of electronic message templates based on the matched member profile attribute values; and

populating at least one electronic message template attribute field with at least one member profile attribute value selected from the matched member profile attribute values.

3. The system of claim 1, wherein the system is configured to match the first plurality of member profile attribute values with the second plurality of member profile attribute values by:

comparing at least one member profile attribute value selected from the first plurality of member profile attribute values with at least one member profile attribute value selected from the second plurality of member profile attribute values.

4. The system of claim 1, wherein the system is configured to match the first plurality of member profile attribute values with the second plurality of member profile attribute values by:

referencing a member profile attribute relationship graph to determine a relationship of at least one member profile attribute value selected from the first plurality of member profile attribute values with at least one member profile attribute value selected from the second plurality of member profile attribute values; and

matching the at least one member profile attribute value selected from the first plurality of member profile attribute values with the at least one member profile attribute value selected from the second plurality of member profile attribute values where there is an edge that connects a first node representing the at least one member profile attribute value selected from the first plurality of member profile attribute values with a second node representing the at least one member profile attribute value selected from the second plurality of member profile attribute values.

5. The system of claim 1, wherein the system is configured to generate the personalized message by:

providing the matched member profile attribute values as an input to a long-short term memory (LSTM) recurrent neural network (RNN); and

wherein

the LSTM RNN generates the personalized message using at least one matched member profile attribute value and a sequence-to-sequence model based on previously sent electronic messages.

6. The system of claim 1, wherein the system is further configured to:

display the personalized message for editing by a member associated with the first member profile; and

receive at least one edit to the personalized message prior to sending the personalized electronic message to the member associated with the selected second member profile.

7. The system of claim 1, wherein the system is further configured to generate the personalized electronic message by:

determining a deviation score for each of the matched member profile attribute values;

ranking each of the matched member profile attribute values based on their corresponding deviation score; and

selecting a subset of the matched member profile attribute values for generating the personalized electronic message based on the rank assigned to each of the matched member profile attribute values and a ranked attribute threshold that defines the rankings to include in the subset of the matched member profile attribute values.

8. A method comprising:

selecting a first member profile of a networked communication service;

determining a plurality of member profiles from the networked communication service based on the selected first member profile;

receiving a selection of a second member profile, the second member profile being selected from the first plurality of member profiles;

receiving an instruction to generate an electronic message for sending to a member of the networked communication service associated with the selected second member profile;

matching a first plurality of member profile attribute values associated with the first member profile with a second plurality of member profile attribute values associated with the second member profile to obtain a plurality of matched member profile attribute values;

generating a personalized electronic message for the member associated with the selected second member profile based on the plurality of matched member profile attribute values; and

sending the personalized electronic message to the member associated with the selected second member profile.

9. The method of claim 8, wherein generating the personalized electronic comprises:

identifying an electronic message template from a plurality of electronic message templates based on the matched member profile attribute values; and

populating at least one electronic message template attribute field with at least one member profile attribute value selected from the matched member profile attribute values.

10. The method of claim 8, wherein matching the first plurality of member profile attribute values with the second plurality of member profile attribute values comprises:

comparing at least one member profile attribute value selected from the first plurality of member profile attribute values with at least one member profile attribute value selected from the second plurality of member profile attribute values.

11. The method of claim 8, wherein matching the first plurality of member profile attribute values with the second plurality of member profile attribute values comprises:

referencing a member profile attribute relationship graph to deter mine a relationship of at least one member profile attribute value selected from the first plurality of member profile attribute values with at least one member profile attribute value selected from the second plurality of member profile attribute values; and

matching the at least one member profile attribute value selected from the first plurality of member profile attribute values with the at least one member profile attribute value selected from the second plurality of member profile attribute values where there is an edge that connects a first node representing the at least one member profile attribute value selected from the first plurality of member profile attribute values with a second node representing the at least one member profile attribute value selected from the second plurality of member profile attribute values.

12. The method of claim 8, wherein generating the personalized message comprises:

providing the matched member profile attribute values as an input to a long-short term memory (LSTM) recurrent neural network (RIND); and

generating, by the LSTM RNN, the personalized message using at least one matched member profile attribute value and a sequence-to-sequence model based on previously sent electronic messages.

13. The method of claim 8, further comprising:

displaying the personalized message for editing by a member associated with the first member profile; and

receiving at least one edit to the personalized message prior to sending the personalized electronic message to the member associated with the selected second member profile.

14. The method of claim 8, wherein generating the personalized electronic message comprises:

determining a deviation score for each of the matched member profile attribute values;

ranking each of the matched member profile attribute values based on their corresponding deviation score; and

selecting a subset of the matched member profile attribute values for generating the personalized electronic message based on the rank assigned to each of the matched member profile attribute values and a ranked attribute threshold that defines the rankings to include in the subset of the matched member profile attribute values.

15. A computer-readable storage device having computer-executable instructions stored thereon that, when executed by one or more hardware processors, causes a system to perform a plurality of operations comprising:

selecting a first member profile of a networked communication service;

determining a plurality of member profiles from the networked communication service based on the selected first member profile;

receiving a selection of a second member profile, the second member profile being selected from the first plurality of member profiles;

receiving an instruction to generate an electronic message for sending to a member of the networked communication service associated with the selected second member profile;

matching a first plurality of member profile attribute values associated with the first member profile with a second plurality of member profile attribute values associated with the second member profile to obtain a plurality of matched member profile attribute values;

generating a personalized electronic message for the member associated with the selected second member profile based on the plurality of matched member profile attribute values; and

sending the personalized electronic message to the member associated with the selected second member profile.

16. The computer-readable storage device of claim 15, generating the personalized electronic message comprises:

identifying an electronic message template from a plurality of electronic message templates based on the matched member profile attribute values; and

populating at least one electronic message template attribute field with at least one member profile attribute value selected from the matched member profile attribute values.

17. The computer-readable storage device of claim 15, wherein matching the first plurality of member profile attribute values with the second plurality of member profile attribute values comprises:

comparing at least one member profile attribute value selected from the first plurality of member profile attribute values with at least one member profile attribute value selected from the second plurality of member profile attribute values.

18. The computer-readable storage device of claim 15, wherein matching the first plurality of member profile attribute values with the second plurality of member profile attribute values comprises:

referencing a member profile attribute relationship graph to determine a relationship of at least one member profile attribute value selected from the first plurality of member profile attribute values with at least one member profile attribute value selected from the second plurality of member profile attribute values; and

matching the at least one member profile attribute value selected from the first plurality of member profile attribute values with the at least one member profile attribute value selected from the second plurality of member profile attribute values where there is an edge that connects a first node representing the at least one member profile attribute value selected from the first plurality of member profile attribute values with a second node representing the at least one member profile attribute value selected from the second plurality of member profile attribute values.

19. The computer-readable storage device of claim 15, wherein generating the personalized message comprises:

providing the matched member profile attribute values as an input to a long-short term memory (LSTM) recurrent neural network (RNN); and

generating, by the LSTM RNN, the personalized message using at least one matched member profile attribute value and a sequence-to-sequence model based on previously sent electronic messages.

20. The computer-readable storage device of claim 15, wherein generating the personalized electronic message comprises:

determining a deviation score for each of the matched member profile attribute values;

ranking each of the matched member profile attribute values based on their corresponding deviation score; and

selecting a subset of the matched member profile attribute values for generating the personalized electronic message based on the rank assigned to each of the matched member profile attribute values and a ranked attribute threshold that defines the rankings to include in the subset of the matched member profile attribute values.