Methods, systems, and apparatus for predicting characteristics of a user
View Patent ↗Methods, systems, and apparatus for predicting the characteristics of a user are described. A model based on a conditional multivariate normal distribution and social relationship information between the selected user and each of one or more other users are obtained. One or more characteristics of the selected user are determined based on the model and the social relationship information. The user characteristics may be determined by adjusting the characteristics of a typical source user according to the model and the social relationship information of the selected user.
1. A method for predicting one or more characteristics of a selected user, comprising:
obtaining a model based on a conditional multivariate normal distribution that relates social relationship information of one or more source users to characteristics of the one or more source users;
obtaining social relationship information relating the selected user and each of one or more other users; and
determining, using a hardware device, one or more characteristics of the selected user based on the model and the social relationship information that relates the selected user and each of the one or more other users,
wherein determining the one or more characteristics comprises:
multiplying an inverted m×m social relationship covariance matrix and a n×m user characteristics social relationship covariance matrix to derive a n×m combined covariance matrix, wherein the social relationship covariance matrix represents a deviation associated with pairs of typical values of the social relationship information of the one or more source users;
subtracting a m×1 social relationship mean vector from a m×1 social relationship input vector to derive a social relationship difference vector, wherein the social relationship mean vector represents the characteristics of the one or more source users;
multiplying the derived n×m combined covariance matrix and the social relationship difference vector to derive a user characteristic adjustment vector; and
adding the user characteristic adjustment vector and a n×1 user characteristics mean vector to derive a n×1 user characteristics vector.
2. The method of claim 1 , wherein the social relationship information comprises one or more counts of hops between the selected user and each of the one or more other users.
3. The method of claim 1 , wherein the determining adjusts one or more characteristics of a typical user, as determined from the one or more source users, to determine the one or more characteristics of the selected user.
4. The method of claim 1 , wherein the model is based on the conditional multivariate normal distribution with mean vector m and covariance matrix s.
5. The method of claim 1 , wherein the model is based on a social relationships mean vector representing a number of hops from a typical source user to each of the one or more source users.
6. The method of claim 1 , wherein the model comprises a m×m social relationship covariance matrix, the m×m social relationship covariance matrix representing a covariance of social relationships of two or more of the one or more source users.
7. The method of claim 1 , wherein the model is based on a n×m user characteristics covariance social relationship matrix, the n×m user characteristics covariance social relationship matrix representing a covariance of one or more characteristics of two or more source users with social relationships of the two or more source users.
8. The method of claim 1 , wherein the model is based on a n×1 user characteristics mean vector, the n×1 user characteristics mean vector representing values of characteristics for a typical source user.
9. The method of claim 1 , wherein the social relationship information comprises a m×1 social relationship input vector, the m×1 social relationship input vector representing social relationships between the selected user and each of the one or more other users.
10. The method of claim 1 , wherein the determining includes determining a n×1 predicted user characteristics vector that represents one or more predicted characteristics of the selected user.
11. The method of claim 1 , further comprising inverting a m×m social relationship covariance matrix to derive the inverted m×m social relationship covariance matrix.
12. An apparatus for predicting one or more characteristics of a selected user, the apparatus comprising:
a processor;
memory to store instructions that, when executed by the processor cause the processor to:
obtain a model based on a conditional multivariate normal distribution that relates social relationship information of one or more source users to characteristics of the one or more source users; obtain social relationship information relating the selected user and each of one or more other users; and
determine one or more characteristics of the selected user based on the model and the social relationship information that relates the selected user and the one or more other users, wherein determining the one or more characteristics comprises:
multiplying an inverted m×m social relationship covariance matrix and a n×m user characteristics social relationship covariance matrix to derive a n×m combined covariance matrix, wherein the social relationship covariance matrix represents a deviation associated with pairs of typical values of the social relationship information of the one or more source users;
subtracting a m×1 social relationship mean vector from a m×1 social relationship input vector to derive a social relationship difference vector, wherein the social relationship mean vector represents the characteristics of the one or more source users;
multiplying the derived n×m combined covariance matrix and the social relationship difference vector to derive a user characteristic adjustment vector; and
adding the user characteristic adjustment vector and a n×1 user characteristics mean vector to derive a n×1 user characteristics vector.
13. The apparatus of claim 12 , wherein the social relationship information comprises one or more counts of hops between the selected user and each of the one or more other users.
14. The apparatus of claim 12 , wherein the determination adjusts one or more characteristics of a typical user, as determined from the one or more source users, to determine the one or more characteristics of the selected user.
15. The apparatus of claim 12 , wherein the model is based on a conditional multivariate normal distribution with mean vector m and covariance matrix s.
16. The apparatus of claim 12 , wherein the model comprises a social relationships mean vector, the social relationships mean vector representing a number of hops from a typical source user to each of the one or more source users.
17. The apparatus of claim 12 , wherein the model comprises a m×m social relationship covariance matrix, the m×m social relationship covariance matrix representing a covariance of social relationships of two or more of the one or more source users.
18. The apparatus of claim 12 , wherein the model is based on a n×m user characteristics covariance social relationship matrix, the n×m user characteristics covariance social relationship matrix representing a covariance of one or more characteristics of two or more source users with social relationships of the two or more of the source users.
19. The apparatus of claim 12 , wherein the model is based on a n×1 user characteristics mean vector, the n×1 user characteristics mean vector representing values of characteristics for a typical source user.
20. The apparatus of claim 12 , wherein the social relationship information comprises a m×1 social relationship input vector, the m×1 social relationship input vector representing social relationships between the selected user and each of the one or more other users.
21. The apparatus of claim 12 , wherein the determining includes determining a n×1 predicted user characteristics vector that represents one or more predicted characteristics of the selected user.
22. The apparatus of claim 12 , wherein the determining the one or more characteristics of the selected user further comprises inverting a m×m social relationship covariance matrix to derive the inverted m×m social relationship covariance matrix.
23. An article of manufacture comprising a non-transitory machine-readable storage medium having machine executable instructions embedded thereon, which when executed by a machine, cause the machine to:
obtain a model based on a conditional multivariate normal distribution that relates social relationship information of one or more source users to characteristics of the one or more source users;
obtain social relationship information relating a selected user and each of one or more other users; and
determine one or more characteristics of the selected user based on the model and the social relationship information that relates the selected user and the one or more other users, wherein determining the one or more characteristics comprises:
multiplying an inverted m×m social relationship covariance matrix and a n×m user characteristics social relationship covariance matrix to derive a n×m combined covariance matrix, wherein the social relationship covariance matrix represents a deviation associated with pairs of typical values of the social relationship information of the one or more source users;
subtracting a m×1 social relationship mean vector from a m×1 social relationship input vector to derive a social relationship difference vector, wherein the social relationship mean vector represents the characteristics of the one or more source users;
multiplying the derived n×m combined covariance matrix and the social relationship difference vector to derive a user characteristic adjustment vector; and
adding the user characteristic adjustment vector and a n×1 user characteristics mean vector to derive a n×1 user characteristics vector.