System and method for automating market analysis from anonymous behavior profiles
First Claim
1. A system for digital media engagement, comprising:
- a web-based computer system and network configured as;
(a) an application client which presents media options as objects to system users, wherein said system users are subjects, and measures and collects affinities of said subjects to said objects as actual affinities A, and further wherein said application client transmits said affinities to a recommendation server across said network;
(b) the recommendation server which represents and stores a profile of a said subject as a subject vector S, and a profile of a said object as an object vector B, such that a predicted affinity P of the subject to the object is generated by matching the subject vector to the object vector according to the dot product of S and B;
(c) a profiler of said recommendation server which trains the subject vectors and object vectors as a minimization of a cost function, E, based on the differences between the predicted and actual affinities of the subjects for the objects, for which actual affinities have been measured and stored in at least one database of the recommendation server, wherein the recommendation server starts with subject and object vectors of a specified number of dimensions, N, which values of the dimensions of the subject and object vectors are initially set to random values;
then the object vectors are fixed and the subject vectors are updated to minimize the cost function, E;
then the subject vectors are fixed and the object vectors are updated to minimize the cost function, E; and
the above steps of fixing the object vectors and updating the subject vectors and then fixing the subject vectors and updating the object vectors are repeated until the change in the decrease of the cost function at the end of the update of the object vectors for that iteration is less than a specified limit, and the resulting trained subject vectors and object vectors are updated in the at least one database of the recommendation server;
whereby said cost function, E, is the mean squared error between the predicted affinities and the actual affinities;
8 Assignments
0 Petitions
Accused Products
Abstract
A system and method is disclosed for profiling subjects and objects based on subjects'"'"' responses to various objects, for purposes of determining and presenting the objects most likely to generate the most likely response from each visitor. Object ratings are explicitly submitted by subjects or derived implicitly from visitor interactions with the objects. A profiling engine processes the ratings information and generates compact profiles of each subject and object based on similarities and differences in affinities between the group of subjects and the group of objects. The object profiles can be clustered to create behavioral object categories. Additionally, a modeling module inverts the abstract subject and object profiles into marketing attributes. The system has application in market analysis and segmentation, behavioral targeting, product placement, and online advertising, to name but a few applications.
-
Citations
17 Claims
-
1. A system for digital media engagement, comprising:
-
a web-based computer system and network configured as; (a) an application client which presents media options as objects to system users, wherein said system users are subjects, and measures and collects affinities of said subjects to said objects as actual affinities A, and further wherein said application client transmits said affinities to a recommendation server across said network; (b) the recommendation server which represents and stores a profile of a said subject as a subject vector S, and a profile of a said object as an object vector B, such that a predicted affinity P of the subject to the object is generated by matching the subject vector to the object vector according to the dot product of S and B; (c) a profiler of said recommendation server which trains the subject vectors and object vectors as a minimization of a cost function, E, based on the differences between the predicted and actual affinities of the subjects for the objects, for which actual affinities have been measured and stored in at least one database of the recommendation server, wherein the recommendation server starts with subject and object vectors of a specified number of dimensions, N, which values of the dimensions of the subject and object vectors are initially set to random values;
then the object vectors are fixed and the subject vectors are updated to minimize the cost function, E;
then the subject vectors are fixed and the object vectors are updated to minimize the cost function, E; and
the above steps of fixing the object vectors and updating the subject vectors and then fixing the subject vectors and updating the object vectors are repeated until the change in the decrease of the cost function at the end of the update of the object vectors for that iteration is less than a specified limit, and the resulting trained subject vectors and object vectors are updated in the at least one database of the recommendation server;whereby said cost function, E, is the mean squared error between the predicted affinities and the actual affinities; - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9)
-
-
10. A system for digital media engagement, comprising:
-
a web-based computer system and network configured to include; (a) an application client which presents media options as objects to system users, wherein said system users are subjects, and measures and collects affinities of said subjects to said objects as actual affinities A, and further wherein said application client transmits said affinities to a recommendation server across said network; (b) the recommendation server which represents and stores a profile of a said subject as a subject vector, S, and a profile of a said object as an object vector, B, such that a predicted affinity, P, of the subject to the object is generated by matching the subject vector to the object vector according to the dot product of S and B; (c) a profiler of said recommendation server which trains the subject vectors and object vectors as a minimization of a cost function E, based on the differences between the predicted and actual affinities of the subjects for the objects for which actual affinities have been measured and stored in at least one database of the recommendation server, the recommendation server starts with subject and object vectors of a specified number of dimensions, N, which values of the dimensions of the subject and object vectors are initially set to random values;
then the object vectors are fixed and the subject vectors are updated to minimize the cost function, E;
then the subject vectors are fixed and the object vectors are updated to minimize the cost function, E; and
the above steps of fixing the object vectors and updating the subject vectors and then fixing the subject vectors and updating the object vectors are repeated until the change in the decrease of the cost function at the end of the update of the object vectors for that iteration is less than a specified limit, and the resulting trained subject vectors and object vectors are updated in the at least one database of the recommendation server;whereby said cost function, E, is the mean squared error between the predicted affinities and the actual affinities; - View Dependent Claims (11, 12, 13, 14, 15, 16, 17)
-
Specification