Systems and methods for classifying content using matrix factorization
View Patent ↗Systems and methods for classifying documents each having zero or more links thereto include generating a link matrix; generating a document term matrix; and jointly factorizing the document term matrix and the link matrix.
1. A computer-implemented method for classifying documents each having zero or more links thereto, comprising:
a. generating a link matrix;
b. generating a document term matrix;
c. jointly factorizing the document term matrix and the link matrix to classify the documents, wherein the link matrix is generated by matrix factorization to derive a feature representation Z of web pages based on analyzing a link matrix A, where said Z is an n×l matrix with each row being an l-dimensional feature vector of a web page;
d. applying a factorization defined as:
min
Z
,
U
A
-
ZUZ
T
F
2
+
γ
U
F
2
+
δ
Z
F
2
where F is a Frobius form, T is a transpose, U is a l×l full matrix, and γ and δ are positive numbers; and
e. determining
min
V
,
Z
C
-
ZV
T
F
2
+
β
V
F
2
where C denotes content of web pages as a matrix whose rows represent documents and columns represent keywords, V denotes a latent space for words, and β is a positive number.
2. The computer-implemented method of claim 1 , comprising generating one or more indicative factors to represent the documents.
3. The computer-implemented method of claim 1 , wherein the document comprises one of: a Web page, an email, and a referenced document.
4. The computer-implemented method of claim 1 , comprising using latent semantic space to represent the documents.
5. The computer-implemented method of claim 4 , comprising mapping the documents to the latent semantic space by determining a link structure and content of the documents.
6. The computer-implemented method of claim 1 , comprising representing the documents as vectors.
7. The computer-implemented method of claim 6 , comprising classifying the vectors.
8. The computer-implemented method of claim 7 , comprising generating one or more classification results.
9. The computer-implemented method of claim 6 , comprising clustering the vectors.
10. The computer-implemented method of claim 9 , comprising generating one or more clustering results.
11. The computer-implemented method of claim 1 , comprising generating a document word co-occurrence matrix from the documents.
12. The computer-implemented method of claim 1 , wherein the link matrix is generated by the matrix factorization to represent web pages as one or more principal factors of a link structure.
13. The computer-implemented method of claim 1 , wherein generating the document term matrix comprises applying a bag-of-words approach.
14. The computer-implemented method of claim 1 , comprising performing supervised matrix factorization.
15. The computer-implemented method of claim 14 , comprising using data labels to guide the matrix factorization.
16. The computer-implemented method of claim 14 , comprising determining a Tikhonov regularization.
17. The computer-implemented method of claim 16 , comprising applying a gradient method.
18. The computer-implemented method of claim 1 , comprising generating a response to a search query based on the jointly factorized document term and link matrices.