System and method for bootstrapping a collaborative filtering system
First Claim
1. A method of predicting a user'"'"'s rating for an item in a collaborative filtering system, comprising:
- providing a correlation coefficient for each pair of users in the system, wherein the correlation coefficient is a measure of the similarity in ratings between pairs of users who have rated a particular item;
determining ratings for items rated by other users in the system;
calculating the weighted average of all the ratings for the item, wherein the weighted average is the sum of the product of a rating and its respective correlation coefficient divided by the sum of the correlation coefficients to provide a predicted user rating;
wherein the plurality of users are members of a predetermined organization; and
wherein the correlation coefficient for each user in the system comprises a predetermined organizational relationship among the users.
8 Assignments
0 Petitions
Accused Products
Abstract
A system and method of predicting a user'"'"'s rating of a new item in a collaborative filtering system in which an initial set of correlation coefficients for the intended users is used to bootstrap the system is described. The users are members of a predetermined organization and the initial correlation coefficient for each pair of users is based on the organizational relationship between the users. Prior organizational relationship information pertaining to the strength of ties, such as a formal organization chart and social network maps built using interviews or deduced from observed (online and/or offline) interaction patterns between potential users, is used to bootstrap the filtering system. Correlation coefficients can be updated as users rate or rerate items in the system.
426 Citations
25 Claims
-
1. A method of predicting a user'"'"'s rating for an item in a collaborative filtering system, comprising:
-
providing a correlation coefficient for each pair of users in the system, wherein the correlation coefficient is a measure of the similarity in ratings between pairs of users who have rated a particular item;
determining ratings for items rated by other users in the system;
calculating the weighted average of all the ratings for the item, wherein the weighted average is the sum of the product of a rating and its respective correlation coefficient divided by the sum of the correlation coefficients to provide a predicted user rating;
wherein the plurality of users are members of a predetermined organization; and
wherein the correlation coefficient for each user in the system comprises a predetermined organizational relationship among the users. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10)
where Siy is the rating of each user Y who has rated the item i, P0 is a predetermined value and α
xy is the correlation coefficient between the user X and the user Y.
-
-
4. The method of claim 3, wherein the initial correlation coefficient for each pair of users X and Y in the system comprises the relationship
-
5. The method of claim 1, further comprising:
providing a map of competencies among the users in the organization.
-
6. The method of claim 1, wherein the correlation coefficient for each pair of users in the system further comprises user specified correlations between pairs of users in the organization.
-
7. The method of claim 1, further comprising:
-
receiving a user rating for the item; and
using the user rating to update the user'"'"'s correlation coefficients with other users.
-
-
8. The method of claim 3, further comprising:
-
updating the correlation α
xy in accordance with the relationship;
are the individual update formulas for user X'"'"'s and user Y'"'"'s ratings distributions, respectively over items rated in common, Txy(0)is the weight attributed to a prior estimate of the user X to user Y correlation, and N is a function of the users X and Y and Txy.
-
-
9. The method of claim 7, further comprising:
for each updated rating of an item rated in common by X and Y, backtracking to remove the effect of the prior rating.
-
10. The method of claim 9, wherein the prior rating pair is removed in accordance with the relationship:
-
are the individual update formulas for user X'"'"'s and user Y'"'"'s ratings distributions, respectively over items rated in common, Txy is the weight attributed to the prior estimate of the user to user correlation, and N is a function of the users and Txy.
-
-
11. A collaborative filtering system for predicting a user'"'"'s rating for an item, comprising:
-
a memory storing;
a correlation coefficient for each pair of users in the system, wherein the correlation coefficient is a measure of the similarity in ratings between pairs of users in the system who have rated at least one item in common; and
ratings for the item made by other users in the system;
a processor for calculating the weighted average of all the ratings for the item, wherein the weighted average is the sum of the product of a rating and its respective correlation coefficient divided by the sum of the correlation coefficients to provide a predicted user rating;
wherein the users are members of a predetermined organization; and
wherein the correlation coefficient for each user in the system comprises a predetermined organizational relationship among the users. - View Dependent Claims (12, 13, 14, 15, 16, 17, 18, 19)
where Siy is the rating of each user y who has rated the item i, P0 is a predetermined value and α
xy is the correlation coefficient between the user X and user Y.
-
-
14. The system of claim 13, wherein the initial correlation coefficient for each pair of users X and Y in the system comprises the relationship
-
15. The system of claim 11, wherein the initial correlation coefficient for each user in the system further comprises user specified correlations between pairs of users in the organization.
-
16. The system of claim 11, wherein the processor, responsive to receiving a user rating for the item, uses the user rating to update the user'"'"'s correlation coefficients.
-
17. The system of claim 16, wherein the processor updates the correlation α
-
xy in accordance with the relationship;
are the individual update formulas for user X'"'"'s and user Y'"'"'s ratings distributions, respectively over items rated in common, Txy is the weight attributed to a prior estimate of the user X to user y correlation, and N is a function of the users X and Y and Txy.
-
xy in accordance with the relationship;
-
18. The system of claim 17, wherein the processor, responsive to an updated rating by a user, calculates a backtrack to remove the effect of the rating on all correlations values that have taken it into account.
-
19. The system of claim 18, wherein the backtrack calculation is made in accordance with the relationship:
-
are the individual update formulas for user'"'"'s X and user'"'"'s Y ratings distributions, respectively over items rated in common, Txy is the weight attributed to the prior estimate of the user to user correlation, and N is a function of the users and Txy.
-
-
20. A method of predicting a user'"'"'s rating for an item in a collaborative filtering system, comprising:
-
providing an initial correlation coefficient for each pair of users in the system;
wherein the plurality of users are members of a predetermined organization; and
wherein the correlation coefficient for each user in the system comprises a predetermined organizational relationship among the users;
determining ratings for items rated by other users in the system; and
calculating the weighted average of all the ratings for the item, wherein the weighted average is the sum of the product of a rating and its respective correlation coefficient divided by the sum of the correlation coefficients to provide a predicted user rating. - View Dependent Claims (21, 22, 23, 24, 25)
-
Specification