Identification of users for advertising using data with missing values

• US 8,676,645 B2
• Filed: 09/12/2012
• Issued: 03/18/2014
• Est. Priority Date: 05/31/2007
• Status: Expired due to Fees

• Current Assignee: Kyndryl Incorporated
• Original Assignee: International Business Machines Corporation
• Inventors: Raghunandan, Hulikunta Prahlad
• Primary Examiner(s): Wu, Rutao
• Assistant Examiner(s): MITTAL, KRISHAN K
• Application Number:
  US13/611,217
• Publication Number:
  US 20130006778A1
• Time in Patent Office: 552 Days
• Field of Search: None
• US Class Current: 705/14.4
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0269 based on user profile or attribute

Abstract

A method and system for identifying users for advertising. Users, attributes, and first web sites provided by ISPs accessed by the users are identified. First data including content of the first web sites and user time spent thereon are received from ISPs and analyzed to determine first attribute values indicative of user interest. Second data received from ISPs include content of second web sites and user time spent thereon. Second attribute values, derived from questionnaires completed by the users, indicate interest in each attribute by each user. Third attribute values are determined by combining the first and second attribute values. The third attribute values are processed to identify users to which a product or service may be advertised. The identified users are communicated to a service provider or product provider. The first, second, or third attribute values may have missing values, which are determined by correlation and linear regression.

Claims

1. A method for identifying users for advertising purposes, said method comprising:

receiving, by a processor of a computer system, from at least one Internet Service Provider (ISP), first data comprising content of web sites of the Internet accessed by N users and time data pertaining to when each user of the N users accessed the web sites, said access to the web sites provided to the N users by the at least one ISP, wherein N≧

4;

said processor analyzing the first data, said analyzing comprising determining first attribute values comprising a first value of each attribute of M attributes for each user, wherein M≧

2, wherein the first value of each attribute is indicative of a level of interest in each attribute by each user, said analyzing based on an amount of time spent by each user at each website of the first web sites in relation to the content of each website of the first web sites, wherein V_n,m,1 denotes the determined first attribute value for attribute m of user n for n=1, 2, . . . , N and m=1, 2, . . . , M;

said processor identifying a second network of second web sites of the Internet accessed by the plurality users, said access to the second web sites provided to the N users by a second at least one ISP;

said processor determining, from questionnaires completed by the N users, second attribute values comprising a second value of each attribute of the plurality of attributes for each user, wherein the second value is indicative of a level of interest in each attribute by each user, wherein attribute values of the second attribute values have been indicated on the questionnaire by the users of the N users for each attribute to which a keyword relevant to the second web sites has been mapped, and wherein second data pertains to said access of the second web sites by the N users, and wherein V_n,m,2 denotes the determined second attribute value for attribute m of user n for n=1, 2, . . . , N and m=1, 2, . . . , M;

said processor determining third attribute values that comprise a third value of each attribute of the plurality of attributes for each user, by combining the first attribute values for each user with the second attribute values for each user,

wherein the first, second, and third attribute values are binary numbers, or the first, second, and third attribute values are real numbers, or the first, second, and third attribute values are positive integers,

wherein said determining V_n,m,k for k=1, k=2, or k=3 comprises determining V_n,m,k for all user-attribute pairs (n,m) of user n and attribute m for n=1, 2, . . . N and m=1, 2, . . . , M except for a single user-attribute pair (n1,m1) such that n1 is 1, 2, . . . , or N and m1 is 1, 2, . . . , or M, followed by performing the steps of;

selecting attribute m2 of the M attributes subject to m2≠

m1, said selecting m2 comprising determining that V_n,m1,k is linearly correlated with V_n,m2,k for a user class consisting of N_UC users, wherein the N_UC users initially consist of the N users minus user n1 such that N_UC=N−

1;

after said selecting attribute m2, performing a linear regression to determine a regression equation expressing V_n,m1,k as a linear function of V_n,m2,k for the user class; and

computing V_n1,m1,1 via the regression equation.

2. The method of claim 1, said method further comprising:

said processor processing the third attribute values, comprising determining from the third attribute values an identification of a subset of the N users to whom advertising of a product or service may be directed; and

said processor communicating the identification of the subset of the N users to a provider of the product or service.

3. The method of claim 1,

wherein for a function F(V_n,m,k) of V_n,m,k for n=1, 2, . . . N and m=1, 2, . . . , M and k=1, 2, said determining third attribute values V_n,m,3 comprises computing V_n,m,3 according to V_n,m,3=W_m,1*F(V_n,m,1)+W_m,2*F(V_n,m,2) such that W_m,k is a weight that acts as a multiplier on F(V_n,m,k) for k=1, 2.

4. The method of claim 1, wherein said selecting attribute m2 comprises:

calculating a Pearson correlation coefficient r(m1,m) between V_n,m1,1 and V_n,m,1 for the user class, for a limited range R of m=1, 2, . . . , M excluding m=m1;

after said calculating r(m1,m) for the limited range R, selecting attribute m2 such that r²(m1,m2) is maximal with respect to all r²(m1,m) in the limited range R;

ascertaining whether a set of conditions are satisfied, wherein the set of conditions are that |r(m1,m2)| is not less than a minimum level of correlation (r_min) and that |r(m1,m2)|/σ

(m1,m2) denoted as t(m1,m2) exceeds a threshold level (t_th), and wherein σ

(m1,m2) is a standard error of r(m1,m2) and is equal to [(1−

r²(m1,m2))/(N_UC−

2)]^1/2,

wherein if said ascertaining ascertains that the set of conditions are not satisfied, then changing the user class and performing said calculating r(m1,m) for the changed user class followed by verifying that the set of conditions are satisfied.

5. The method of claim 4, wherein said ascertaining ascertains that the set of conditions are not satisfied, and wherein said changing the user class comprises:

(i) adding X new users to the user class which increments N_UG by X such that X≧

1;

or (ii) removing Y users from the user class which decrements N_UG by Y such that Y≧

1 subject to N_UG being at least 3 after being decremented by Y.

6. The method of claim 5, wherein the threshold level t_th is a critical value (t_crit) of at distribution having N_UC−

2 degrees of freedom at a specified significance level α

, and wherein t(m1,m2) exceeding t_crit rejects a null hypothesis of no linear correlation between attribute m1 and attribute m2.

7. The method of claim 6,

wherein the method further comprises prior to said calculating r(m1,m) for the limited range R, specifying a first significance level α

_i and a second significance level α

₂, such that α

₂≠

α

₁; and

wherein said selecting attribute m2 comprises imposing a constraint of α

=α

_i if r(m1,m2)>

0 and α

=α

₂ if r(m1,m2)<

0.

8. A computer program product, comprising a computer readable hardware storage device having a computer readable program code embodied therein, said computer readable program code containing instructions that when executed by a processor of a computer system perform a method for identifying users for advertising purposes, said method comprising:

said processor receiving from at least one Internet Service Provider (ISP), first data comprising content of web sites of the Internet accessed by N users and time data pertaining to when each user of the N users accessed the web sites, said access to the web sites provided to the N users by the at least one ISP, wherein N≧

4;

said processor analyzing the first data, said analyzing comprising determining first attribute values comprising a first value of each attribute of M attributes for each user, wherein M≧

2, wherein the first value of each attribute is indicative of a level of interest in each attribute by each user, said analyzing based on an amount of time spent by each user at each website of the first web sites in relation to the content of each website of the first web sites, wherein V_n,m,1 denotes the determined first attribute value for attribute m of user n for n=1, 2, . . . , N and m=1, 2, . . . , M;

said processor identifying a second network of second web sites of the Internet accessed by the plurality users, said access to the second web sites provided to the N users by a second at least one ISP;

said processor determining, from questionnaires completed by the N users, second attribute values comprising a second value of each attribute of the plurality of attributes for each user, wherein the second value is indicative of a level of interest in each attribute by each user, wherein attribute values of the second attribute values have been indicated on the questionnaire by the users of the N users for each attribute to which a keyword relevant to the second web sites has been mapped, and wherein second data pertains to said access of the second web sites by the N users, and wherein V_n,m,2 denotes the determined second attribute value for attribute m of user n for n=1, 2, . . . , N and m=1, 2, . . . , M;

said processor determining third attribute values that comprise a third value of each attribute of the plurality of attributes for each user, by combining the first attribute values for each user with the second attribute values for each user,

wherein the first, second, and third attribute values are binary numbers, or the first, second, and third attribute values are real numbers, or the first, second, and third attribute values are positive integers,

wherein said determining V_n,m,k for k=1, k=2, or k=3 comprises determining V_n,m,k for all user-attribute pairs (n,m) of user n and attribute m for n=1, 2, . . . N and m=1, 2, . . . , M except for a single user-attribute pair (n1,m1) such that n1 is 1, 2, . . . , or N and m1 is 1, 2, . . . , or M, followed by performing the steps of;

selecting attribute m2 of the M attributes subject to m2≠

m1, said selecting m2 comprising that V_n,m1,k is linearly correlated with V_n,m2,k for a user class consisting of N_UC users, wherein the N_UC users initially consist of the N users minus user n1 such that N_UC=N−

1;

after said selecting attribute m2, performing a linear regression to determine a regression equation expressing V_n,m1,k as a linear function of V_n,m2,k for the user class; and

computing V_n1,m1,1 via the regression equation.

9. The computer program product of claim 8, said method further comprising:

said processor processing the third attribute values, comprising determining from the third attribute values an identification of a subset of the N users to whom advertising of a product or service may be directed; and

said processor communicating the identification of the subset of the N users to a provider of the product or service.

10. The computer program product of claim 8,

wherein for a function F(V_n,m,k) of V_n,m,k for n=1, 2, . . . N and m=1, 2, . . . , M and k=1, 2, said determining third attribute values V_n,m,3 comprises computing V_n,m,3 according to V_n,m,3=W_m,1*F(V_n,m,1)+W_m,2*F(V_n,m,2) such that W_m,k is a weight that acts as a multiplier on F(V_n,m,k) for k=1, 2.

11. A computer system comprising a processor, a memory coupled to the processor, and a computer readable storage device coupled to the processor, said storage device containing program code configured to be executed by the processor via the memory to implement a method for identifying users for advertising purposes, said method comprising:

said processor receiving from at least one Internet Service Provider (ISP), first data comprising content of web sites of the Internet accessed by N users and time data pertaining to when each user of the N users accessed the web sites, said access to the web sites provided to the N users by the at least one ISP, wherein N≧

4;

said processor analyzing the first data, said analyzing comprising determining first attribute values comprising a first value of each attribute of M attributes for each user, wherein M≧

2, wherein the first value of each attribute is indicative of a level of interest in each attribute by each user, said analyzing based on an amount of time spent by each user at each website of the first web sites in relation to the content of each website of the first web sites, wherein V_n,m1,1 denotes the determined first attribute value for attribute m of user n for n=1, 2, . . . , N and m=1, 2, . . . , M;

said processor identifying a second network of second web sites of the Internet accessed by the plurality users, said access to the second web sites provided to the N users by a second at least one ISP;

said processor determining, from questionnaires completed by the N users, second attribute values comprising a second value of each attribute of the plurality of attributes for each user, wherein the second value is indicative of a level of interest in each attribute by each user, wherein attribute values of the second attribute values have been indicated on the questionnaire by the users of the N users for each attribute to which a keyword relevant to the second web sites has been mapped, and wherein second data pertains to said access of the second web sites by the N users, and wherein V_n,m,2 denotes the determined second attribute value for attribute m of user n for n=1, 2, . . . , N and m=1, 2, . . . , M;

said processor determining third attribute values that comprise a third value of each attribute of the plurality of attributes for each user, by combining the first attribute values for each user with the second attribute values for each user,

wherein the first, second, and third attribute values are binary numbers, or the first, second, and third attribute values are real numbers, or the first, second, and third attribute values are positive integers,

wherein said determining V_n,m,k for k=1, k=2, or k=3 comprises determining V_n,m,k for all user-attribute pairs (n,m) of user n and attribute m for n=1, 2, . . . N and m=1, 2, . . . , M except for a single user-attribute pair (n1,m1) such that n1 is 1, 2, . . . , or N and m1 is 1, 2, . . . , or M, followed by performing the steps of;

selecting attribute m2 of the M attributes subject to m2≠

m1, said selecting m2 comprising determining that V_n,m1,k is linearly correlated with V_n,m2,k for a user class consisting of N_UC users, wherein the N_UC users initially consist of the N users minus user n1 such that N_UC=N−

1;

after said selecting attribute m2, performing a linear regression to determine a regression equation expressing V_n,m1,k as a linear function of V_n,m2,k for the user class; and

computing in V_n1,m1,1 via the regression equation.

12. The computer system of claim 11, said method further comprising:

said processor processing the third attribute values, comprising determining from the third attribute values an identification of a subset of the N users to whom advertising of a product or service may be directed; and

said processor communicating the identification of the subset of the N users to a provider of the product or service.

13. The computer system of claim 11,

wherein for a function F(V_n,m,k) of V_n,m,k for n=1, 2, . . . N and m=1, 2, . . . , M and k=1, 2, said determining third attribute values V_n,m,3 comprises computing V_n,m,3 according to V_n,m,3=W_m,1*F(V_n,m,1) W_m,2*F(V_n,m,2) such that W_m,k is a weight that acts as a multiplier on F(V_n,m,k) for k=1, 2.

14. A process for supporting computer infrastructure, said process comprising providing at least one support service for at least one of creating, integrating, hosting, maintaining, and deploying computer-readable code in a computer system that comprises a processor and a computer readable storage device storing the code, wherein the code in combination with the computer system is configured to perform, via execution of the code by the processor, a method for identifying users for advertising purposes, said method comprising:

receiving, by a processor of the computer system, from at least one Internet Service Provider (ISP), first data comprising content of web sites of the Internet accessed by N users and time data pertaining to when each user of the N users accessed the web sites, said access to the web sites provided to the N users by the at least one ISP, wherein N≧

4;

said processor analyzing the first data, said analyzing comprising determining first attribute values comprising a first value of each attribute of M attributes for each user, wherein M≧

2, wherein the first value of each attribute is indicative of a level of interest in each attribute by each user, said analyzing based on an amount of time spent by each user at each website of the first web sites in relation to the content of each website of the first web sites, wherein V_n,m,1 denotes the determined first attribute value for attribute m of user n for n=1, 2, . . . , N and m=1, 2, . . . , M;

said processor identifying a second network of second web sites of the Internet accessed by the plurality users, said access to the second web sites provided to the N users by a second at least one ISP;

said processor determining, from questionnaires completed by the N users, second attribute values comprising a second value of each attribute of the plurality of attributes for each user, wherein the second value is indicative of a level of interest in each attribute by each user, wherein attribute values of the second attribute values have been indicated on the questionnaire by the users of the N users for each attribute to which a keyword relevant to the second web sites has been mapped, and wherein second data pertains to said access of the second web sites by the N users, and wherein V_n,m,2 denotes the determined second attribute value for attribute m of user n for n=1, 2, . . . , N and m=1, 2, . . . , M;

said processor determining third attribute values that comprise a third value of each attribute of the plurality of attributes for each user, by combining the first attribute values for each user with the second attribute values for each user,

wherein the first, second, and third attribute values are binary numbers, or the first, second, and third attribute values are real numbers, or the first, second, and third attribute values are positive integers,

wherein said determining V_n,m,k for k=1, k=2, or k=3 comprises determining V_n,m,k for all user-attribute pairs (n,m) of user n and attribute m for n=1, 2, . . . N and m=1, 2, . . . , M except for a single user-attribute pair (n1,m1) such that n1 is 1, 2, . . . , or N and m1 is 1, 2, . . . , or M, followed by performing the steps of;

selecting attribute m2 of the M attributes subject to m2≠

m1, said selecting m2 comprising determining that V_n,m1,k is linearly correlated with V_n,m2,k for a user class consisting of N_UC users, wherein the N_UC users initially consist of the N users minus user n1 such that N_UC=N−

1;

after said selecting attribute m2, performing a linear regression to determine a regression equation expressing V_n,m1,k as a linear function of V_n,m2,k for the user class; and

computing in V_n1,m1,1 via the regression equation.

15. The process of claim 14, said method further comprising:

said processor processing the third attribute values, comprising determining from the third attribute values an identification of a subset of the N users to whom advertising of a product or service may be directed; and

said processor communicating the identification of the subset of the N users to a provider of the product or service.

16. The process of claim 14,

wherein for a function F(V_n,m,k) of V_n,m,k for n=1, 2, . . . N and m=1, 2, . . . , M and k=1, 2, said determining third attribute values V_n,m,3 comprises computing V_n,m,3 according to V_n,m,3=W_m,1*F(V_n,m,1)+W_m,2*F(V_n,m,2) such that W_m,k is a weight that acts as a multiplier on F(V_n,m,k) for k=1, 2.