Determining brand affinity of users
A method including a user who has brand affinities about a product brand or a product brand category. The method can include analyzing whether the user has brand affinities for the product brand or the brand category with two binary classification models that can include (a) a model one to determine an affinity to a particular product brand and (b) a model two to determine an affinity to a particular product brand category. The method also can include analyzing whether the user has affinities for a similar product brand or a similar product brand category by a model three neural network. The method can include displaying recommendations and the promotions for a product brand, a product brand category, a similar product brand, or a similar product brand category to the user. Other embodiments are disclosed.
1. A system comprising:
one or more processors; and
one or more non-transitory computer-readable media storing computing instructions that, when executed on the one or more processors cause the one or more processors to perform functions comprising:
identifying a first user who has brand affinities about a first product brand or a first product brand category by using two models, wherein:
(a) a model one determines an affinity to a particular first product brand, wherein when a first result of the model one exceeds a first predetermined threshold value for a product brand within a predetermined period of time, the first result is a first probability score; and
(b) a model two determines an affinity to a particular first product brand category;
analyzing whether the first user has affinities for a similar product brand or a similar product brand category by using a model three neural network to determine an affinity for the similar product brand or the similar product brand category, wherein the similar product brand is similar to the first product brand, and wherein the similar product brand category is similar to the first product brand category;
determining whether to display first recommendations for the first product brand or the first product brand category to the first user based on results from the two models;
determining whether to display second recommendations for the similar product brand or the similar product brand category to the first user based on output from the model three neural network; and
transmitting instructions to display the first and second recommendations to the first user.
2. The system of claim 1 , wherein the computing instructions, when executed on the one or more processors, further cause the one or more processors to perform a function comprising:
retrieving online purchase histories and online browser histories of the first user from one or more databases corresponding to the first product brand and the first product brand category, wherein the online purchase histories comprise purchase transactions, and wherein the online browser histories comprise histories of web browsing and histories of viewing brand specific webpages.
3. The system of claim 1 , wherein, the model one to determine the affinity to the particular first product brand further comprises the first user represented as:
Affinity
for
Brand
M
=
P
(
Brand
=
M
❘
"\[LeftBracketingBar]"
User
Features
for
Brand
=
M
)
=
P
(
B
=
M
❘
"\[LeftBracketingBar]"
F
M
1
,
F
M
2
F
M
3
,
⋯
)
wherein, P is for probability, B is a brand, M is a particular brand, F is a feature of the first user and F M1 , F M2 , F M3 , . . . are features related to a user for whom brand affinity is being computed.
4. The system of claim 1 , wherein the model two to determine the affinity to the particular first product brand category further comprises a binary classification model similar to the model one except that positive labels for the binary classification model uses only affinity to a brand for a given brand category while treating other brands as negative labels.
5. The system of claim 1 , wherein the two models further comprises using two binary classification models, wherein:
a given vector of features for the first user is used to classify whether the first user has an affinity for a given brand; and
the two binary classification models are trained using the given vector of features associated with one or more activities of the first user as training input for the two binary classification models.
6. The system of claim 5 , wherein multiple regression models are trained to derive affinity of the first user to the given brand in a given product brand category.
7. The system of claim 6 , wherein when a second result of the model two exceeds a second predetermined threshold value for a product brand category within the predetermined period of time, the second result is a second probability score.
8. The system of claim 1 , wherein analyzing whether the first user has the affinities for the similar product brand or the similar product brand category by using the model three neural network further comprises using a cosine similarity model to identify top K brands out of all brands for which the first user has an affinity, where the top K brands can be controlled through thresholds determined by the first user.
9. The system of claim 1 , wherein the model three neural network further comprises:
identification of similar brands by a neural network model with item attributes and first user features comprising co-bought, co-clicked, and co-viewed brands to obtain latent vector representation of product brands;
extracting the product brands from online purchase histories or online browser histories of the first user, wherein the product brands are treated as words in a sentence as part of the neural network model; and
generating word embedding from the sentence to infer latent features to find product similarities.
10. The system of claim 1 , wherein using the model three neural network to determine an affinity for the similar product brand or the similar product brand category further comprises determining affinities of the first user for one or more similar product brands or one or more similar product brand categories in a given product category corresponding to a first product in one or more first product categories.
11. A method being implemented via execution of computing instructions configured to run on one or more processors and stored on one or more non-transitory computer-readable media, the method comprising:
one or more processors; and
one or more non-transitory computer-readable media storing computing instructions that, when executed on the one or more processors cause the one or more processors to perform functions comprising:
identifying a first user who has brand affinities about a first product brand or a first product brand category by using two models, wherein:
(a) a model one determines an affinity to a particular first product brand, wherein when a first result of the model one exceeds a first predetermined threshold value for a product brand within a predetermined period of time, the first result is a first probability score; and
(b) a model two determines an affinity to a particular first product brand category;
analyzing whether the first user has affinities for a similar product brand or a similar product brand category by using a model three neural network to determine an affinity for the similar product brand or the similar product brand category, wherein the similar product brand is similar to the first product brand, and wherein the similar product brand category is similar to the first product brand category;
determining whether to display first recommendations for the first product brand or the first product brand category to the first user based on results from the two models;
determining whether to display second recommendations for the similar product brand or the similar product brand category to the first user based on output from the model three neural network; and
transmitting instructions to display the first and second recommendations to the first user and promotions to the first user.
12. The method of claim 11 , further comprising:
retrieving online purchase histories and online browser histories of the first user from one or more databases corresponding to the first product brand and the first product brand category, wherein the online purchase histories comprise purchase transactions, and wherein the online browser histories comprise histories of web browsing and histories of viewing brand specific webpages.
13. The method of claim 11 , wherein, the model one to determine the affinity to the particular first product brand further comprises the first user represented as:
Affinity
for
Brand
M
=
P
(
Brand
=
M
❘
"\[LeftBracketingBar]"
User
Features
for
Brand
=
M
)
=
P
(
B
=
M
❘
"\[LeftBracketingBar]"
F
M
1
,
F
M
2
F
M
3
,
⋯
)
wherein, P is for probability, B is a brand, M is a particular brand, F is a feature of the first user and F M1 , F M2 , F M3 , . . . are features related to a user for whom brand affinity is being computed.
14. The method of claim 11 , wherein the model two to determine the affinity to the particular first product brand category further comprises a binary classification model similar to the model one except that positive labels for the binary classification model uses only affinity to a brand for a given brand category while treating other brands as negative labels.
15. The method of claim 11 , wherein the two models further comprises using two binary classification models, wherein:
a given vector of features for the first user is used to classify whether the first user has an affinity for a given brand; and
the two binary classification models are trained using the given vector of features associated with one or more activities of the first user as training input for the two binary classification models.
16. The method of claim 15 , wherein multiple regression models are trained to derive affinity of the first user to the given brand in a given product brand category.
17. The method of claim 16 , wherein when a second result of the model two exceeds a second predetermined threshold value for a product brand category within the predetermined period of time, the second result is a second probability score.
18. The method of claim 11 , wherein analyzing whether the first user has the affinities for the similar product brand or the similar product brand category by using the model three neural network further comprises using a cosine similarity model to identify top K brands out of all brands for which the first user has an affinity, where the top K brands can be controlled through thresholds determined by the first user.
19. The method of claim 11 , wherein the model three neural network further comprises:
identification of similar brands by a neural network model with item attributes and first user features comprising co-bought, co-clicked, and co-viewed brands to obtain latent vector representation of product brands;
extracting the product brands from online purchase histories or online browser histories of the first user, wherein the product brands are treated as words in a sentence as part of the neural network model; and
generating word embedding from the sentence to infer latent features to find product similarities.
20. The method of claim 11 , wherein using the model three neural network to determine an affinity for the similar product brand or the similar product brand category further comprises determining affinities of the first user for one or more similar product brands or one or more similar product brand categories in a given product category corresponding to a first product in one or more first product categories.