Computerized system and method for multi-factor message classification and delivery
Disclosed are systems and methods for improving interactions with and between computers in content providing, searching and/or hosting systems supported by or configured with devices, servers and/or platforms. The disclosed systems and methods provide a novel messaging framework that automatically applies a multi-factor analysis technique to incoming and received messages in order to properly identify a message's type and category, which dictates the manner in which the message is displayed within a recipient's inbox. The disclosed framework operates on two levels: i) it determines whether a message is from a human or machine sender (H/M classification), and ii) it determines the messages category (MAGMA categorization).
1. A method comprising:
receiving, by a device over a network, a message sent by a sender that is addressed to an inbox of a recipient, the message comprising message content;
parsing, by the device, the message, and identifying message information related to the sender, the message content and a subject of the message;
applying, by the device, a Human/Machine (H/M) classifier to the message, the H/M classifier comprising at least one sub-model that analyzes a message based on specific features of the message, the application of the H/M classifier comprising analyzing the sender information and the content information via at least a content model, such that application of the content model comprises:
creating a vector representation of each word in the sender information;
creating a vector representation of each word in the content information;
analyzing, via a CNN layer of the content model, each vector, and based on the analysis, concatenate each vector to a content model vector; and
analyzing the content model vector via a softmax function, and based on the analysis, determining whether the message sender was a machine or a human;
identifying, by the device, based on the H/M classifier application and analysis, information indicating whether the message sender is a machine or human;
applying, by the device, a MAGMA classifier to the message, the MAGMA classifier comprising a word encoder for specific portions of a message, the application of the MAGMA classifier comprising analyzing the subject information and the content information via the portion specific word encoder;
determining, by the device, based on the MAGMA classifier application, information indicating at least one specific category the message is associated with;
applying, by the device, a set of indicators to the message, the set of indicators providing information related to the machine or human determination and the MAGMA categorization; and
communicating, by the device, over the network, the message to the inbox along with the set of indicators.
2. The method of claim 1 , wherein the H/M classifier further comprises a sender model, action model and salutation model.
3. The method of claim 2 , wherein application of the sender model comprises:
identifying, from the sender information, an email address and a name of the sender;
generating a vector representation of the email address;
generating a vector representation of the name;
concatenating each vector based on convolutional analysis of each vector to generate a sender model vector; and
analyzing the sender model vector, and based on the analysis, determining whether the message sender was a machine or a human.
4. The method of claim 2 , wherein application of the action model comprises:
identifying data indicating actions related to message deletions within the inbox;
identifying data related to actions of opened messages within the inbox;
training a CNN model based on the identified deletion and opened data;
applying the trained CNN model to the sender information and content information; and
determining whether the message sender was a machine or a human.
5. The method of claim 2 , wherein application of the salutation model comprises:
identifying data indicating a positive label, the positive label corresponding to messages that contain an explicit salutation, the explicit salutation being a message with a match between the n beginning part of a message body and a recipient name;
identifying data indicating a negative label, the negative label corresponding to messages that do not contain the explicit salutation;
training a CNN model based on the identified positive label and negative label data;
applying the trained CNN model to at least a portion of the message content; and
determining whether the message sender was a machine or a human.
6. The method of claim 1 , wherein the at least one specific category is selected from a group of MAGMA classes comprising: Human (personal), Social, Travel, Career, Shopping, Finance and Other.
7. The method of claim 1 , wherein the communication causes the message to be displayed in a portion of a user interface of the inbox based on at least one of the set of indicators.
8. The method of claim 1 , further comprising:
searching, over a network, a datastore associated with third party entities based on information associated with the message;
identifying, based on the search, third party digital content from the datastore, the third party digital content; and
communicating, over the network, along with the communicated message for display within the inbox, the identified third party digital content.
9. A non-transitory computer-readable storage medium tangibly encoded with computer-executable instructions, that when executed by a device, perform a method comprising:
receiving, by the device over a network, a message sent by a sender that is addressed to an inbox of a recipient, the message comprising message content;
parsing, by the device, the message, and identifying message information related to the sender, the message content and a subject of the message;
applying, by the device, a Human/Machine (H/M) classifier to the message, the H/M classifier comprising at least one sub-model that analyzes a message based on specific features of the message, the application of the H/M classifier comprising analyzing the sender information and the content information via at least a content model, such that application of the content model comprises:
creating a vector representation of each word in the sender information;
creating a vector representation of each word in the content information;
analyzing, via a CNN layer of the content model, each vector, and based on the analysis, concatenate each vector to a content model vector; and
analyzing the content model vector via a softmax function, and based on the analysis, determining whether the message sender was a machine or a human;
identifying, by the device, based on the H/M classifier application and analysis, information indicating whether the message sender is a machine or human;
applying, by the device, a MAGMA classifier to the message, the MAGMA classifier comprising a word encoder for specific portions of a message, the application of the MAGMA classifier comprising analyzing the subject information and the content information via the portion specific word encoder;
determining, by the device, based on the MAGMA classifier application, information indicating at least one specific category the message is associated with;
applying, by the device, a set of indicators to the message, the set of indicators providing information related to the machine or human determination and the MAGMA categorization; and
communicating, by the device, over the network, the message to the inbox along with the set of indicators.
10. The non-transitory computer-readable storage medium of claim 9 , wherein the H/M classifier further comprises a sender model, action model and salutation model.
11. The non-transitory computer-readable storage medium of claim 10 , wherein application of the sender model comprises:
identifying, from the sender information, an email address and a name of the sender;
generating a vector representation of the email address;
generating a vector representation of the name;
concatenating each vector based on convolutional analysis of each vector to generate a sender model vector; and
analyzing the sender model vector, and based on the analysis, determining whether the message sender was a machine or a human.
12. The non-transitory computer-readable storage medium of claim 10 , wherein application of the action model comprises:
identifying data indicating actions related to message deletions within the inbox;
identifying data related to actions of opened messages within the inbox;
training a CNN model based on the identified deletion and opened data;
applying the trained CNN model to the sender information and content information; and
determining whether the message sender was a machine or a human.
13. The non-transitory computer-readable storage medium of claim 10 , wherein application of the salutation model comprises:
identifying data indicating a positive label, the positive label corresponding to messages that contain an explicit salutation, the explicit salutation being a message with a match between the n beginning part of a message body and a recipient name;
identifying data indicating a negative label, the negative label corresponding to messages that do not contain the explicit salutation;
training a CNN model based on the identified positive label and negative label data;
applying the trained CNN model to at least a portion of the message content; and
determining whether the message sender was a machine or a human.
14. The non-transitory computer-readable storage medium of claim 9 , wherein the at least one specific category is selected from a group of MAGMA classes comprising: Human (personal), Social, Travel, Career, Shopping, Finance and Other.
15. The non-transitory computer-readable storage medium of claim 9 , wherein the communication causes the message to be displayed in a portion of a user interface of the inbox based on at least one of the set of indicators.
16. A device comprising:
a processor configured to:
receive, over a network, a message sent by a sender that is addressed to an inbox of a recipient, the message comprising message content;
parse the message, and identify message information related to the sender, the message content and a subject of the message;
apply a Human/Machine (H/M) classifier to the message, the H/M classifier comprising at least one sub-model that analyzes a message based on specific features of the message, the application of the H/M classifier comprising analyzing the sender information and the content information via at least a content model, such that application of the content model comprises:
create a vector representation of each word in the sender information;
create a vector representation of each word in the content information;
analyze, via a CNN layer of the content model, each vector, and based on the analysis, concatenate each vector to a content model vector; and
analyze the content model vector via a softmax function, and based on the analysis, determine whether the message sender was a machine or a human;
identify, based on the H/M classifier application and analysis, information indicating whether the message sender is a machine or human;
apply a MAGMA classifier to the message, the MAGMA classifier comprising a word encoder for specific portions of a message, the application of the MAGMA classifier comprising analyzing the subject information and the content information via the portion specific word encoder;
determine, based on the MAGMA classifier application, information indicating at least one specific category the message is associated with;
apply a set of indicators to the message, the set of indicators providing information related to the machine or human determination and the MAGMA categorization; and
communicate, over the network, the message to the inbox along with the set of indicators.
17. The device of claim 16 , wherein the H/M classifier further comprises a sender model, action model and salutation model.
18. The device of claim 17 , wherein application of the sender model comprises:
identify, from the sender information, an email address and a name of the sender;
generate a vector representation of the email address;
generate a vector representation of the name;
concatenate each vector based on convolutional analysis of each vector to generate a sender model vector; and
analyze the sender model vector, and based on the analysis, determine whether the message sender was a machine or a human.
19. The device of claim 17 , wherein application of the action model comprises:
identify data indicating actions related to message deletions within the inbox;
identify data related to actions of opened messages within the inbox;
train a CNN model based on the identified deletion and opened data;
apply the trained CNN model to the sender information and content information; and
determine whether the message sender was a machine or a human.
20. The device of claim 17 , wherein application of the salutation model comprises:
identify data indicating a positive label, the positive label corresponding to messages that contain an explicit salutation, the explicit salutation being a message with a match between the n beginning part of a message body and a recipient name;
identify data indicating a negative label, the negative label corresponding to messages that do not contain the explicit salutation;
train a CNN model based on the identified positive label and negative label data;
apply the trained CNN model to at least a portion of the message content; and
determine whether the message sender was a machine or a human.