US-12620007-B2 - Consumer sentiment analysis for selection of creative elements

US12620007B2US 12620007 B2US12620007 B2US 12620007B2US-12620007-B2

Abstract

The subject technology predicts consumer sentiment based on demographics and other static features of the consumer as well as dynamic features generated based on engagement of the consumer with previously presented targeted content. The sentiment predictions are used to recommend and generate new targeted content that is published to the consumer.

Inventors

Steven Gerber
Pavan Korada
David Schey
Sunpreet Khanuja
Gayan De Silva

Assignees

ZETA GLOBAL CORP.

Dates

Publication Date: 20260505
Application Date: 20240724

Claims (20)

1 . A digital publication system configured to publish targeted content to online consumers, the digital publication system comprising: a memory storing machine executable code; and one or more processors being connected to the memory and one or more datasets associated with the online consumers via a publication network, the one or more processors being configurable to execute the machine executable code so that the processor is configured to: receive one or more raw events for one or more exposure time periods, the one or more raw events being received in response to a delivery of one or more pieces of targeted content each comprising at least one universal tag to one or more internet enabled devices through one or more vendor servers over the one or more exposure time periods, the one or more pieces of targeted content being associated with one or more sentiments, the at least one universal tag activating a script in response to the one or more pieces of targeted content being accessed by the one or more internet enabled devices; receive a cookie associated with each of the one or more internet enabled devices in response to the script being activated by the at least one universal tag; determine a set of static features based on the received cookie and a set of dynamic features based on the received one or more raw events; train a sentiment prediction model based on the set of static features and the set of dynamic features, the sentiment prediction model being configured to determine one or more engagement probabilities for the one or more sentiments; determine a creative element associated with one of the one or more sentiments having an increased engagement probability based on the one or more engagement probabilities determined by the sentiment prediction model; and cause a display of the creative element on an internet enabled device via the one or more vendor servers.
2 . The system of claim 1 , wherein the one or more processors are further configured to manipulate at least one of the one or more exposure time periods to generate an overlapping exposure time period for at least one of the one or more internet enabled devices, the overlapping exposure time period being generated by extending a defined period of time for the at least one manipulated exposure time period to capture a delivery of multiple pieces of targeted content to the least one of the one or more internet enabled devices; determine a training sample for the sentiment prediction model, the training sample including one or more dynamic features determined based on at least one raw event recorded during the overlapping exposure time period; and train the sentiment prediction model based on the training sample.
3 . The system of claim 1 , wherein the one or more processors are further configured to determine an identity and one or more exposure windows associated with the one or more internet enabled devices, the one or more exposure windows including an identifier for a piece of targeted content presented to the one or more internet enabled devices, one or more sentiment tags for at least one creative element being included in the piece of targeted content, a digital timestamp recording a time the piece of targeted content was presented to the one or more internet enabled devices, a defined time period indicating a temporal length of the one or more exposure windows, and a piece of impression data being captured during the defined time period.
4 . The system of claim 1 , wherein the set of static features includes at least one of a piece of location data or at least one intender attribute for each of the one or more internet enabled devices.
5 . The system of claim 1 , wherein the one or more processors are further configured to derive one or more dynamic features from impression data included in one or more exposure windows.
6 . The system of claim 1 , wherein the one or more pieces of targeted content includes an email; and the one or more processors are further configured to publish the one or more pieces of targeted content by delivering the email to an email server hosting an email inbox of at least one of the one or more internet enabled devices.
7 . The system of claim 1 , wherein the one or more pieces of targeted content includes a display advertisement; and the one or more processors are further configured to publish the one or more pieces of targeted content by providing the display advertisement to a web server hosting a web page including one or more placements for advertising content.
8 . The system of claim 1 , wherein the one or more processors are further configured to select, from one or more exposure windows, a reduced sample of exposure windows including a piece of targeted content having a particular sentiment; and train the sentiment prediction model based on a set of dynamic features derived from the reduced sample of exposure windows.
9 . A method of publishing targeted content to online consumers comprising: receiving one or more raw events for one or more exposure time periods, the one or more raw events being received in response to a delivery of one or more pieces of targeted content each comprising at least one universal tag to one or more internet enabled devices through one or more vendor servers over the one or more exposure time periods, the one or more pieces of targeted content being associated with one or more sentiments, the at least one universal tag activating a script in response to the one or more pieces of targeted content being accessed by the one or more internet enabled devices; receiving a cookie associated with each of the one or more internet enabled devices in response to the script being activated by the at least one universal tag; determining a set of static features based on the received cookie and a set of dynamic features based on the received one or more raw events; training a sentiment prediction model based on the set of static features and the set of dynamic features, the sentiment prediction model being configured to determine one or more engagement probabilities for the one or more sentiments; determining a creative element associated with one of the one or more sentiments having an increased engagement probability based on the one or more engagement probabilities determined by the sentiment prediction model; and causing a display of the creative element on an internet enabled device via the one or more vendor servers.
10 . The method of claim 9 , further comprising manipulating at least one of the one or more exposure time periods to generate an overlapping exposure time period for at least one of the one or more internet enabled devices, the overlapping exposure time period being generated by extending a defined period of time for the at least one manipulated exposure time period to capture a delivery of multiple pieces of targeted content to the least one of the one or more internet enabled devices; determining a training sample for the sentiment prediction model, the training sample including one or more dynamic features determined based on at least one raw event recorded during the overlapping exposure time period; and training the sentiment prediction model based on the training sample.
11 . The method of claim 9 , further comprising determining an identity and one or more exposure windows associated with the one or more internet enabled devices, the one or more exposure windows including an identifier for a piece of targeted content presented to the one or more internet enabled devices, one or more sentiment tags for at least one creative element being included in the piece of targeted content, a digital timestamp recording a time the piece of targeted content was presented to the one or more internet enabled devices, a defined time period indicating a temporal length of the one or more exposure windows, and a piece of impression data being captured during the defined time period.
12 . The method of claim 9 , wherein the set of static features includes at least one of a piece of location data or at least one intender attribute for each of the one or more internet enabled devices.
13 . The method of claim 9 , wherein the method further comprises deriving one or more dynamic features from impression data included in one or more exposure windows.
14 . The method of claim 9 , wherein the one or more pieces of targeted content includes an email; and the method further comprises publishing the one or more pieces of targeted content by delivering the email to an email server hosting an email inbox of at least one of the one or more internet enabled devices.
15 . The method of claim 9 , wherein the one or more pieces of targeted content includes a display advertisement; and the method further comprises publishing the one or more pieces of targeted content by providing the display advertisement to a web server hosting a web page including one or more placements for advertising content.
16 . The method of claim 9 , further comprising selecting, from one or more exposure windows, a reduced sample of exposure windows including a piece of targeted content having a particular sentiment; and training the sentiment prediction model based on a set of dynamic features derived from the reduced sample of exposure windows.
17 . A non-transitory machine-readable medium comprising instructions which, when read by a machine, cause the machine to perform operations comprising: receiving one or more raw events for one or more exposure time periods, the one or more raw events received in response to a delivery of one or more pieces of targeted content each comprising at least one universal tag to one or more internet enabled devices through one or more vendor servers over the one or more exposure time periods, the one or more pieces of targeted content being associated with one or more sentiments, the at least one universal tag activating a script in response to the one or more pieces of targeted content being accessed by the one or more internet enabled devices; receiving a cookie associated with each of the one or more internet enabled devices in response to the script being activated by the at least one universal tag; determining a set of static features based on the received cookie and a set of dynamic features based on the received one or more raw events; training a sentiment prediction model based on the set of static features and the set of dynamic features, the sentiment prediction model being configured to determine one or more engagement probabilities for the one or more sentiments; determining a creative element associated with one of the one or more sentiments having an increased engagement probability based on the one or more engagement probabilities determined by the sentiment prediction model; and causing a display of the creative element on an internet enabled device via the one or more vendor servers.
18 . The non-transitory machine-readable medium of claim 17 , wherein the operations further comprise manipulating at least one of the one or more exposure time periods to generate an overlapping exposure time period for at least one of the one or more internet enabled devices, the overlapping exposure time period being generated by extending a defined period of time for the at least one manipulated exposure time period to capture a delivery of multiple pieces of targeted content to the least one of the one or more internet enabled devices; determining a training sample for the sentiment prediction model, the training sample including one or more dynamic features determined based on at least one raw event recorded during the overlapping exposure time period; and training the sentiment prediction model based on the training sample.
19 . The non-transitory machine-readable medium of claim 17 , wherein the operations further comprise determining an identity and one or more exposure windows associated with the one or more internet enabled devices, the one or more exposure windows including an identifier for a piece of targeted content presented to the one or more internet enabled devices, one or more sentiment tags for at least one creative element being included in the piece of targeted content, a digital timestamp recording a time the piece of targeted content was presented to the one or more internet enabled devices, a defined time period indicating a temporal length of the one or more exposure windows, and a piece of impression data being captured during the defined time period.
20 . The non-transitory machine-readable medium of claim 17 , wherein the operations further comprise selecting, from one or more exposure windows, a reduced sample of exposure windows including a piece of targeted content having a particular sentiment; and training the sentiment prediction model based on a set of dynamic features derived from the reduced sample of exposure windows.

Description

CLAIM OF PRIORITY This patent application is a continuation of U.S. patent application Ser. No. 17/735,093, filed May 2, 2022, which application claims the benefit of priority, under 35 U.S.C. Section 119 (e), to Korada et al, U.S. Provisional Patent Application Ser. No. 63/182,553, entitled “SYSTEM AND METHOD OF PREDICTING SENTIMENT SCORES FOR CONSUMERS,” filed on Apr. 30, 2021, which are hereby incorporated by reference herein in their entireties. TECHNICAL FIELD The subject matter disclosed herein generally relates to the technical field of (online) conversion, visitor profiling, and visitor segmenting. More particularly, the disclosure relates the technical field of predicting behavior that can result in conversion online, and that of customizing the content of distributed messages to the particular interest of an individual consumer and the sentiments of that consumer. BACKGROUND As more environments, including digital environments, are becoming more complex, individuals are turning towards various forms of technology for assistance. The same is true for the digital marketing environment. All types of technology support are utilized in the various areas of digital marketing, including, but not limited to, search marketing, display marketing, online advertisement, lead generation, voucher distribution, content personalization and the like. However, there is not a single technology designated to operate across the various areas of digital marketing in order to assist individuals to reach intelligent decisions in order to increase a predefined action (e.g., conversion, downloads, and the like). Therefore, there is a need for a system or method to predict the probability of a particular consumer performing a certain action using a score-driven approach. BRIEF DESCRIPTION OF THE DRAWINGS Some embodiments are illustrated by way of example and not limitation in the figures of the accompanying drawings. FIG. 1 is a schematic representation of a scoring probability system, according to various embodiments of the present disclosure. FIG. 2 is a block diagram of an internet enabled device of the system of FIG. 1, according to various embodiments of the present disclosure. FIG. 3 is a block diagram of the vendor server of FIG. 1, according to various embodiments of the present disclosure. FIG. 4 is a block diagram of the scoring probability server of FIG. 1, according to various embodiments of the present disclosure. FIG. 5 is a block diagram of components of the scoring probability server of FIG. 4, according to various embodiments of the present disclosure. FIG. 6 is a schematic flow diagram of a method performed by the system of FIG. 1, according to various embodiments of the present disclosure. FIG. 7 illustrates a timeline including multiple exposure windows, according to various embodiments of the present disclosure. FIG. 8 is a flow diagram of a method performed by the system of FIG. 1, according to various embodiments of the present disclosure. FIG. 9 is a flow diagram of a method performed by the system of FIG. 1, according to various embodiments of the present disclosure. FIG. 10 is a block diagram illustrating a high-level network architecture for recommending and publishing content based on consumer sentiment predictions, according to various embodiments of the present disclosure. FIG. 11 is a block diagram showing architectural aspects of a targeted content publication system, according to various embodiments of the present disclosure. FIG. 12 is a flow chart depicting operations in a method, according to various embodiments of the present disclosure. DETAILED DESCRIPTION The description that follows includes systems, methods, techniques, instruction sequences, and computing machine program products that embody illustrative embodiments of the disclosure. In the following description, for the purposes of explanation, numerous specific details are set forth in order to provide an understanding of various embodiments of the inventive subject matter. It will be evident, however, to those skilled in the art, that embodiments of the inventive subject matter may be practiced without these specific details. In general, well-known instruction instances, protocols, structures, and techniques are not necessarily shown in detail. Described herein are systems and methods for consumer sentiment analysis for real time selection of creative elements. In various embodiments, the system may determine the probability of a particular consumer converting (i.e., making a purchase), clicking a piece of actionable content, or committing a certain action based on static features and/or dynamic features associated with the consumer. The dynamic features may be derived from one or more exposure windows that occurred during a defined exposure period. The exposure windows may include an instance where a consumer was presented with a piece of targeted content and impression data recording engagement (e.g., displays, views, clicks,