On-line recommender system

• US 7,475,027 B2
• Filed: 02/06/2003
• Issued: 01/06/2009
• Est. Priority Date: 02/06/2003
• Status: Expired due to Fees

• Current Assignee: Mitsubishi Electric Research Laboratories, Inc. (Mitsubishi Electric Corporation)
• Original Assignee: Mitsubishi Electric Research Laboratories, Inc. (Mitsubishi Electric Corporation)
• Inventors: Brand, Matthew E.
• Primary Examiner(s): Zeender; F.
• Assistant Examiner(s): SHAAWAT, MUSSA A
• Application Number:
  US10/361,165
• Publication Number:
  US 20040158497A1
• Time in Patent Office: 2,161 Days
• Field of Search: None
• US Class Current: 705/7.29
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0201 Market modelling; Market analysis; Collecting market data
• CPC Class Codes: G06Q 30/0253 During e-commerce, i.e. online transactions
• CPC Class Codes: G06Q 30/0631 Item recommendations

Abstract

A method and system recommends items such as products and services to consumers. Rating on items are received from consumers as a sequential stream of data. A thin singular value decomposition is updated, one rating at the time, while receiving the ratings. A prediction of a recommendations of particular items for a particular consumer is based on the updated singular value decomposition while receiving the ratings and updating the singular value decomposition. The ratings are discarded after the updating so that a size, structure, and content of an underlying preference matrix is unknown.

Claims

1. A computer implemented method for recommending items to consumers in a recommender system, comprising the steps of:

receiving a stream of rating on items from consumers via a network;

updating sequentially, by a computer, a singular value decomposition of a preference matrix using the stream of ratings, in which each rating is processed one at a time and independent of any other ratings, while receiving the ratings, and wherein the updating is a sequential rank-1 update of the singular value decomposition to produce a thin SVD, and wherein the preference matrix is X, and wherein the singular value decomposition factors the preference matrix X into two orthogonal matrices U and V, and a diagonal matrix S≐

diag(s), such that USV^T=X, and U^TXV=S, where ^T represents the SVD transform, the elements of s are singular values and columns of U and V are left and right singular vectors, respectively, and further comprising;

arranging non-negative elements on the diagonal of S in descending order, and a first non-zero element in each column of U is positive so that the thin SVD is unique; and

discarding all but a predetermined number of the r largest singular values and the corresponding singular vectors so that a product of the resulting thinned matrices, U′

S′

V′

≈

X, is a best rank-r approximation of X in a least-squares sense, and a matrix U′

^TX=S′

V′

^T;

predicting recommendations of particular items for a particular consumer based on the updated singular value decomposition while receiving the ratings and updating the singular value decomposition; and

displaying the recommendations for the particular consumer on a graphical consumer interface via the network.

2. The method of claim 1 further comprising:

receiving the stream of ratings asynchronously in a random order.

3. The method of claim 1 wherein the updating further comprises:

adding, changing, and retracting the ratings from the singular value decomposition of the preference matrix.

4. The method of claim 1 further comprising:

discarding the ratings after the updating so that a size, structure, and content of the underlying preference matrix is unknown.

5. The method of claim 1 wherein a projection of X onto r orthogonal axes, specified by the columns of U′

is a reduce-dimension representation of the ratings.

6. The method of claim 5 wherein the preference matrix X represents a tabulation of the ratings, an a subspace spanned by the columns of U′

is interpreted as an r-dimensional consumer taste space.

7. The method of claim 6 wherein individual consumers are located in the taste space according to a similarity of their ratings, and a relationship between the ratings of the particular consumer is represented as a column vector c, and a corresponding taste-space location p is c=U′

p, and p=U′

^Tc.

8. The method of claim 6 wherein a product taste space is the vector V′

, and the taste space contains items arranged according to the consumers.

9. The method of claim 1 further comprising:

translating the ratings to center the ratings on an origin of the singular value decomposition.

10. The method of claim 9 further comprising:

interpreting singular value decomposition as a Gaussian covariance model of the ratings that express correlations between the ratings of the consumers.

11. The method of claim 1 where for column vectors a and b, and the thin SVD USV^T=X, the updating determines the singular value decomposition of X+ab^T, where column b is a binary vector indicating which columns are updated, and column a is derived from the ratings.

12. The method of claim 1 wherein all truncated singular values have a small predetermined threshold value ε

.

13. The method of claim 1 further comprising:

displaying the recommendation in real-time on the graphical consumer interface while receiving and updating.

14. The method of claim 1 wherein the receiving, updating, and predicting are executed in a client computer.

15. The method of claim 14 further comprising:

executing the receiving, updating, and predicting in a plurality of client computers connected via a network to enable a distributed and decentralized recommender system.

16. The method of claim 1 wherein the receiving, updating and predicting are all performed on-line, and in real-time.