Granted Patent B2

US 12,248,507 · App. 17/578,838 · Granted Mar 11, 2025

Inventors: Zijia Wang (WeiFang, CN); Jiacheng Ni (Shanghai, CN); Jinpeng Liu (Shanghai, CN); Zhen Jia (Shanghai, CN)

Assignee: Dell Products L.P.

G06F16/435G06F16/24578

Abstract

Claims (99)

1. A method for content recommendation, comprising:

determining, in at least one processing device of an Internet platform, a similarity between a first recommendation result and a second recommendation result for a content set, the first recommendation result and the second recommendation result being determined based on different recommendation techniques and respectively indicative of a recommendation degree for each of a plurality of content items in the content set, wherein the similarity is computed in the form of at least one vector of values indexed by identifiers of respective ones of the content items in the content set;

adjusting, in the at least one processing device of the Internet platform, the second recommendation result using the similarity;

determining, in the at least one processing device of the Internet platform, a target recommendation result for the content set based on the first recommendation result and the adjusted second recommendation result;

providing the target recommendation result from the at least one processing device of the Internet platform to a user device over a network; and

responsive to user interaction with a display presenting the target recommendation result at the user device, providing access to corresponding content at least in part via the at least one processing device of the Internet platform;

wherein determining the similarity between the first and second recommendation results comprises:

determining a first vector comprising a first set of indications based on the first recommendation result, the first vector being indexed by the identifiers of respective ones of the content items in the content set;

determining a second vector comprising a second set of indications based on the second recommendation result, the second vector being indexed by the identifiers of respective ones of the content items in the content set; and

determining the similarity based at least in part on a distance measure computed between the first and second vectors;

wherein adjusting the second recommendation result using the similarity comprises:

determining a modification factor based on the similarity; and

adjusting the second recommendation result based on the modification factor to alter an impact of the second recommendation result in determining the target recommendation result;

wherein adjusting the second recommendation result based on the modification factor comprises computing updated values for respective ones of the second set of indications of the second vector as a function of previous values for the respective ones of the second set of indications and the modification factor.

2. The method according to claim 1 , wherein determining the similarity comprises:

determining the first vector as the first set of indications based on the first recommendation result, the first set of indications being respectively indicative of recommendation degrees for corresponding contents in the content set;

determining the second vector as the second set of indications based on the second recommendation result, the second set of indications being respectively indicative of recommendation degrees for corresponding contents in the content set; and

determining the distance measure between the first and second vectors at least in part by computing a cosine similarity between the first set of indications and the second set of indications.

3. The method according to claim 1 , wherein adjusting the second recommendation result comprises:

determining a scaling factor based on the similarity; and

scaling the second recommendation result using the scaling factor.

4. The method according to claim 1 , wherein determining the target recommendation result comprises:

acquiring a first optimized recommendation result corresponding to the first recommendation result using the first recommendation result and the adjusted second recommendation result; and

determining the target recommendation result based on the first optimized recommendation result.

5. The method according to claim 4 , wherein determining the target recommendation result based on the first optimized recommendation result comprises:

adjusting the first recommendation result using the similarity;

acquiring a second optimized recommendation result corresponding to the second recommendation result using the second recommendation result and the adjusted first recommendation result; and

determining the target recommendation result based on the first optimized recommendation result and the second optimized recommendation result.

6. The method according to claim 1 , wherein the content set is a set of candidate articles and the recommendation technique recommends a candidate article from the set of candidate articles based on a predetermined base article, and the method further comprises:

determining a set of correlation measures associated with the set of candidate articles, each correlation measure in the set of correlation measures being indicative of a correlation degree between a corresponding candidate article and the base article; and

adjusting the target recommendation result using the set of correlation measures, so as to update the target recommendation result.

7. The method according to claim 6 , wherein the set of candidate articles comprises a first candidate article, the set of correlation measures comprises a first correlation measure corresponding to the first candidate article, and determining the set of correlation measures comprises:

determining, using a pre-trained language model, a correlation degree between a reference of the first candidate article and a title of the base article to serve as the first correlation measure.

8. An electronic device of an Internet platform, the electronic device comprising:

a processor; and

a memory coupled to the processor, wherein the memory has instructions stored therein, and the instructions, when executed by the processor, cause the electronic device to execute actions comprising:

determining a similarity between a first recommendation result and a second recommendation result for a content set, the first recommendation result and the second recommendation result being determined based on different recommendation techniques and respectively indicative of a recommendation degree for each of a plurality of content items in the content set, wherein the similarity is computed in the form of at least one vector of values indexed by identifiers of respective ones of the content items in the content set;

adjusting the second recommendation result using the similarity;

determining a target recommendation result for the content set based on the first recommendation result and the adjusted second recommendation result;

providing the target recommendation result from the Internet platform to a user device over a network; and

responsive to user interaction with a display presenting the target recommendation result at the user device, providing access to corresponding content at least in part via the Internet platform;

wherein determining the similarity between the first and second recommendation results comprises:

determining a first vector comprising a first set of indications based on the first recommendation result, the first vector being indexed by the identifiers of respective ones of the content items in the content set;

determining a second vector comprising a second set of indications based on the second recommendation result, the second vector being indexed by the identifiers of respective ones of the content items in the content set; and

determining the similarity based at least in part on a distance measure computed between the first and second vectors;

wherein adjusting the second recommendation result using the similarity comprises:

determining a modification factor based on the similarity; and

adjusting the second recommendation result based on the modification factor to alter an impact of the second recommendation result in determining the target recommendation result;

wherein adjusting the second recommendation result based on the modification factor comprises computing updated values for respective ones of the second set of indications of the second vector as a function of previous values for the respective ones of the second set of indications and the modification factor.

9. The electronic device according to claim 8 , wherein determining the similarity comprises:

determining the first vector as the first set of indications based on the first recommendation result, the first set of indications being respectively indicative of recommendation degrees for corresponding contents in the content set;

determining the second vector as the second set of indications based on the second recommendation result, the second set of indications being respectively indicative of recommendation degrees for corresponding contents in the content set; and

determining the distance measure between the first and second vectors at least in part by computing a cosine similarity between the first set of indications and the second set of indications.

10. The electronic device according to claim 8 , wherein adjusting the second recommendation result comprises:

determining a scaling factor based on the similarity; and

scaling the second recommendation result using the scaling factor.

11. The electronic device according to claim 8 , wherein determining the target recommendation result comprises:

acquiring a first optimized recommendation result corresponding to the first recommendation result using the first recommendation result and the adjusted second recommendation result; and

determining the target recommendation result based on the first optimized recommendation result.

12. The electronic device according to claim 11 , wherein determining the target recommendation result based on the first optimized recommendation result comprises:

adjusting the first recommendation result using the similarity;

acquiring a second optimized recommendation result corresponding to the second recommendation result using the second recommendation result and the adjusted first recommendation result; and

determining the target recommendation result based on the first optimized recommendation result and the second optimized recommendation result.

13. The electronic device according to claim 8 , wherein the content set is a set of candidate articles, the recommendation technique recommends a candidate article from the set of candidate articles based on a predetermined base article, and the actions further comprise:

determining a set of correlation measures associated with the set of candidate articles, each correlation measure in the set of correlation measures being indicative of a correlation degree between a corresponding candidate article and the base article; and

adjusting the target recommendation result using the set of correlation measures, so as to update the target recommendation result.

14. The electronic device according to claim 13 , wherein the set of candidate articles comprises a first candidate article, the set of correlation measures comprises a first correlation measure corresponding to the first candidate article, and determining the set of correlation measures comprises:

determining, using a pre-trained language model, a correlation degree between a reference of the first candidate article and a title of the base article to serve as the first correlation measure.

15. A computer program product tangibly stored on a computer-readable medium and comprising machine-executable instructions, wherein the machine-executable instructions, when executed by at least one processing device of an Internet platform, cause the at least one processing device of the Internet platform to perform a method for content recommendation, the method comprising:

determining a similarity between a first recommendation result and a second recommendation result for a content set, the first recommendation result and the second recommendation result being determined based on different recommendation techniques and respectively indicative of a recommendation degree for each of a plurality of content items in the content set, wherein the similarity is computed in the form of at least one vector of values indexed by identifiers of respective ones of the content items in the content set;

adjusting the second recommendation result using the similarity;

determining a target recommendation result for the content set based on the first recommendation result and the adjusted second recommendation result;

providing the target recommendation result from the at least one processing device of the Internet platform to a user device over a network; and

responsive to user interaction with a display presenting the target recommendation result at the user device, providing access to corresponding content at least in part via the at least one processing device of the Internet platform;

wherein determining the similarity between the first and second recommendation results comprises:

determining a first vector comprising a first set of indications based on the first recommendation result, the first vector being indexed by the identifiers of respective ones of the content items in the content set;

determining a second vector comprising a second set of indications based on the second recommendation result, the second vector being indexed by the identifiers of respective ones of the content items in the content set; and

determining the similarity based at least in part on a distance measure computed between the first and second vectors;

wherein adjusting the second recommendation result using the similarity comprises:

determining a modification factor based on the similarity; and

adjusting the second recommendation result based on the modification factor to alter an impact of the second recommendation result in determining the target recommendation result;

wherein adjusting the second recommendation result based on the modification factor comprises computing updated values for respective ones of the second set of indications of the second vector as a function of previous values for the respective ones of the second set of indications and the modification factor.

16. The computer program product according to claim 15 , wherein determining the similarity comprises:

determining the first vector as the first set of indications based on the first recommendation result, the first set of indications being respectively indicative of recommendation degrees for corresponding contents in the content set;

determining the second vector as the second set of indications based on the second recommendation result, the second set of indications being respectively indicative of recommendation degrees for corresponding contents in the content set; and

determining the distance measure between the first and second vectors at least in part by computing a cosine similarity between the first set of indications and the second set of indications.

17. The computer program product according to claim 15 , wherein adjusting the second recommendation result comprises:

determining a scaling factor based on the similarity; and

scaling the second recommendation result using the scaling factor.

18. The computer program product according to claim 15 , wherein determining the target recommendation result comprises:

acquiring a first optimized recommendation result corresponding to the first recommendation result using the first recommendation result and the adjusted second recommendation result; and

determining the target recommendation result based on the first optimized recommendation result.

19. The computer program product according to claim 18 , wherein determining the target recommendation result based on the first optimized recommendation result comprises:

adjusting the first recommendation result using the similarity;

acquiring a second optimized recommendation result corresponding to the second recommendation result using the second recommendation result and the adjusted first recommendation result; and

determining the target recommendation result based on the first optimized recommendation result and the second optimized recommendation result.

20. The computer program product according to claim 15 , wherein the content set is a set of candidate articles and the recommendation technique recommends a candidate article from the set of candidate articles based on a predetermined base article, and the method further comprises:

determining a set of correlation measures associated with the set of candidate articles, each correlation measure in the set of correlation measures being indicative of a correlation degree between a corresponding candidate article and the base article; and

adjusting the target recommendation result using the set of correlation measures, so as to update the target recommendation result.

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (47)