Systems and methods for whole page personalization

1. A system comprising:

one or more processors; and

one or more non-transitory computer-readable media storing computing instructions configured to run on the one or more processors and perform:

modeling a webpage as a random field, wherein the random field comprises an undirected graph comprising:

two or more nodes; and

one or more edges,

wherein:

each node of the two or more nodes comprises:

 one or more placements on the webpage; and

 a goodness function of one or more goodness functions associated with one or more webpage elements; and

each edge of the one or more edges comprises a compatibility function based at least in part on the one or more goodness functions of two different nodes of the two or more nodes associated with each edge of the one or more edges;

determining a probability of the webpage having exceeded a predetermined threshold based at least in part on one or more of the compatibility functions of the one or more edges, wherein the one or more compatibility functions comprise (i) a product of a compatibility of a first placement of a first node of the two or more nodes and a second placement of a second node of the two or more nodes, and (ii) the compatibility function of the one or more compatibility functions of an edge of the one or more edges between the first node at the first placement and the second node at the second placement, wherein the one or more placements comprise the first and second placements; and

sending instructions to display the webpage based at least in part on the probability of the webpage having exceeded the predetermined threshold.

2. The system of claim 1 , wherein:

the one or more webpage elements further comprise at least one of:

an advertisement;

a search recommendation; or

an item recommendation.

3. The system of claim 1 , wherein:

the goodness function further comprises a probabilistic model learned from historical session data to generate positive samples and negative samples.

4. The system of claim 1 , wherein:

the compatibility function of each of the one or more edges provides a compatibility score based at least in part on an engagement value, the engagement value comprising at least one of:

a click rate,

an add to cart rate, or

a purchase rate.

5. The system of claim 1 , wherein the computing instructions are further configured perform:

receiving an affinity score of a user for each of one or more categories, the one or more webpage elements comprising the one or more categories.

6. The system of claim 1 , wherein:

determining the probability of the webpage having exceeded the predetermined threshold is further based at least in part on a search query of a user.

7. The system of claim 1 , wherein:

determining the probability of the webpage having exceeded the predetermined threshold is further based at least in part on a user profile of a user.

8. The system of claim 1 , wherein:

the one or more placements comprise contiguous placements of the one or more webpage elements on the webpage.

9. The system of claim 1 , wherein the computing instructions are further configured to perform:

determining the first placement of the one or more placements of the first node of a first webpage element of the one or more webpage elements; and

determining the second placement of the one or more placements of the second node of a second webpage element of the one or more webpage elements.

10. The system of claim 9 , wherein:

determining the probability of the webpage having exceeded the predetermined threshold is determined using a formula:

P

⁡

(

page

)

=

1

Z

⁢

∏

i

⁢

∅

i

⁡

(

x

)

⁢

∏

i

,

j

;

Edge

⁡

(

i

,

j

)

⁢

φ

i

,

j

⁡

(

x

,

y

)

,

wherein:

P(page) is the probability of the webpage having exceeded the predetermined threshold;

Z is a normalization constant;

i is a first node of the two or more nodes;

j is a second node of the two or more nodes;

x is the first placement of the first node;

Ø i (x) is the goodness function of the one or more goodness functions of the first node at the first placement;

y is the second placement of the one or more placements of the second node; and

φ i,j (x, y) is the compatibility function of an edge of the one or more edges between the first node at the first placement and the second node at the second placement.

11. A method comprising:

modeling a webpage as a random field, wherein the random field comprises an undirected graph comprising:

two or more nodes; and

one or more edges,

wherein:

each node of the two or more nodes comprises:

one or more placements on the webpage; and

a goodness function of one or more goodness functions associated with one or more webpage elements; and

each edge of the one or more edges comprises a compatibility function based at least in part on the one or more goodness functions of two different nodes of the two or more nodes associated with each edge of the one or more edges;

determining a probability of the webpage having exceeded a predetermined threshold based at least in part on one or more of the compatibility functions of the one or more edges, wherein the one or more compatibility functions comprise (i) a product of a compatibility of a first placement of a first node of the two or more nodes and a second placement of a second node of the two or more nodes, and (ii) the compatibility function of the one or more compatibility functions of an edge of the one or more edges between the first node at the first placement and the second node at the second placement, wherein the one or more placements comprise the first and second placements; and

sending instructions to display the webpage based at least in part on the probability of the webpage having exceeded the predetermined threshold.

12. The method of claim 11 , wherein:

the one or more webpage elements further comprises at least one of:

an advertisement;

a search recommendation; or

an item recommendation.

13. The method of claim 11 , wherein:

the goodness function further comprises a probabilistic model learned from historical session data to generate positive samples and negative samples.

14. The method of claim 11 , wherein:

the compatibility function of each of the one or more edges provides a compatibility score based at least in part on an engagement value, the engagement value comprising at least one of:

a click rate,

an add to cart rate, or

a purchase rate.

15. The method of claim 11 , further comprising:

receiving an affinity score of a user for each of one or more categories, the one or more webpage elements comprising the one or more categories.

16. The method of claim 11 , wherein:

determining the probability of the webpage having exceeded the predetermined threshold is further based at least in part on a search query of a user.

17. The method of claim 11 , wherein:

determining the probability of the webpage having exceeded the predetermined threshold is further based at least in part on a user profile of a user.

18. The method of claim 11 , wherein:

the one or more placements comprise contiguous placements of the one or more webpage elements on the webpage.

19. The method of claim 11 , further comprising:

determining the first placement of the one or more placements of the first node of a first webpage element of the one or more webpage elements; and

determining the second placement of the one or more placements of the second node of a second webpage element of the one or more webpage elements.

20. The method of claim 19 , wherein:

determining the probability of the webpage having exceeded the predetermined threshold is determined using a formula:

P

⁡

(

page

)

=

1

Z

⁢

∏

i

⁢

∅

i

⁡

(

x

)

⁢

∏

i

,

j

;

Edge

⁡

(

i

,

j

)

⁢

φ

i

,

j

⁡

(

x

,

y

)

,

wherein:

P(page) is the probability of the webpage having exceeded the predetermined threshold;

Z is a normalization constant;

i is a first node of the two or more nodes;

j is a second node of the two or more nodes;

x is the first placement of the first node;

Ø i (x) is the goodness function of the one or more goodness functions of the first node at the first placement;

y is the second placement of the one or more placements of the second node; and

φ i,j (x, y) is the compatibility function of an edge of the one or more edges between the first node at the first placement and the second node at the second placement.