Recommending magazines to users of a digital magazine server

• US 10,311,476 B2
• Filed: 01/24/2014
• Issued: 06/04/2019
• Est. Priority Date: 01/24/2014
• Status: Active Grant

• Current Assignee: Flipboard Incorporated
• Original Assignee: Flipboard Incorporated
• Inventors: Aleksandrovsky, Boris Lev
• Primary Examiner(s): Burke, Jeff A
• Assistant Examiner(s): Dewan, Kamal K
• Application Number:
  US14/164,089
• Publication Number:
  US 20150213492A1
• Time in Patent Office: 1,957 Days
• Field of Search: 707748, 707706, 707749, 707E17009, 707E17002, 707E17003, 707E17005, 707736, 707778, 707723
• CPC Class Codes: G06F 16/3322 using system suggestions G06F16/3325 takes precedence
• CPC Class Codes: G06F 16/335 Filtering based on additional data, e.g. user or group profiles filtering in web context G06F16/9535, G06F16/9536
• CPC Class Codes: G06F 16/355 Class or cluster creation or modification
• CPC Class Codes: G06F 16/435 Filtering based on additional data, e.g. user or group profiles
• CPC Class Codes: G06Q 30/0255 based on user history

Abstract

A digital magazine server identifies content items for recommendation to a user based on content items with which the user previously interacted. Based on key phrases and terms in content items with which the user previously interacted, topics are associated with the content items and used to generate a vector for each content item. The vectors are used to generate clusters including one or more content items. A characteristic vector is generated for each cluster based on the vectors generated for content items within a cluster. Candidate content items are retrieved and topics included in the candidate content items are used along with the characteristic vectors to determine a measure of similarity between candidate content items and various clusters. Candidate content items with at least a threshold measure of similarity to a cluster are selected for presentation to the user.

Claims

1. A computer-implemented method for recommending content items to a user of a digital magazine server, the method comprising:

identifying a specific type of content item for recommending to the user of the digital magazine server;

determining a time window corresponding to the identified specific type of content item, the duration of the time window determined based on a number of content items having the specific type included in a content store of the digital magazine server, different types of content item corresponding to time windows having different durations;

identifying a plurality of content items with which the user of the digital magazine server previously interacted during the determined time window corresponding to the identified specific type of content item;

identifying key terms from each of the plurality of content items;

generating one or more topics associated with each content item based at least in part on the identified key terms and characteristics describing frequency of presentation of each of the identified key terms in each content item and describing differences in presentation of each of the identified key terms by each content item relative to presentation of other terms in each content item;

generating a vector for each content item based at least in part on one or more topics associated with a content item, the vector having one or more dimensions with a value of a dimension based at least in part on a number of times a topic occurs in the content item;

generating one or more clusters each including one or more content items based at least in part on the generated vectors;

determining characteristic vectors for each cluster based at least in part on the generated vectors, a characteristic vector for a cluster based at least in part on one or more of the generated vectors included in the cluster;

retrieving a candidate content item having the specific type;

determining a measure of similarity between the candidate content item and one or more of the characteristic vectors; and

selecting the candidate content item for recommendation to the user based at least in part on the determined measures of similarity.

2. The computer-implemented method of claim 1, wherein identifying the plurality of content items with which the user of the digital magazine server previously interacted comprises:

identifying a plurality of content items having at least a threshold number of connections to the user of the digital magazine server user via the digital magazine server.

3. The computer-implemented method of claim 1, wherein identifying the plurality of content items with which the user of the digital magazine server previously interacted comprises:

identifying a plurality of content items having at least a threshold number of connections to additional content items with which the user of the digital magazine server previously interacted.

4. The computer-implemented method of claim 1, wherein generating the vector for each content item based on one or more topics associated with the content item comprises:

associating a weight with one or more dimensions of the vector based at least in part on a type of the content item for which the vector was generated.

5. The computer-implemented method of claim 1, wherein determining characteristic vectors for each cluster based at least in part on the generated vectors comprises:

determining values associated with one or more dimensions of each vector generated for content items in a cluster;

determining weights associated with one or more dimensions of each vector generated for content items in a cluster; and

generating a characteristic vector for the cluster based at least in part on the determined values and the determined weights.

6. The computer-implemented method of claim 1, wherein determining the measure of similarity between the candidate content item and one or more of the characteristic vectors comprises:

generating a vector for the candidate content item based at least in part on one or more topics associated with the candidate content item, the vector for the candidate content item having one or more dimensions with a value of a dimension based at least in part on a number of times a topic occurs in the candidate content item; and

determining a measure of similarity between the vector generated for the candidate content item and a characteristic vector based at least in part on values of one or more dimensions of the vector generated for the candidate content item and values of one or more dimensions of the characteristic vector.

7. The computer-implemented method of claim 1, wherein selecting the candidate content item for recommendation to the user based at least in part on the determined measures of similarity comprises:

generating a ranking of a plurality of candidate content items based at least in part on measures of similarity between each of the candidate content items and one or more characteristic vectors; and

selecting the candidate content item for recommendation if the candidate content item has at least a threshold position in the ranking.

8. The computer-implemented method of claim 1, wherein selecting the candidate content item for recommendation to the user based at least in part on the determined measures of similarity comprises:

selecting the candidate content item if a measure of similarity between the candidate content item and at least one characteristic vector equals or exceeds a threshold value.

9. A computer-implemented method for recommending content items to a user of a digital magazine server, the method comprising:

identifying a specific type of content item for recommending to the user of the digital magazine server;

determining a time window corresponding to the identified specific type of content item, the duration of the time window determined based on a number of content items having the specific type included in a content store of the digital magazine server, different types of content item corresponding to time windows having different durations;

identifying a plurality of content items with which the user of the digital magazine server previously interacted during the determined time window corresponding to the identified specific type of content item;

generating a vector for each content item based at least in part on one or more topics determined from key terms included in a content item, the vector having one or more dimensions with a value of a dimension based at least in part on a number of times a topic occurs in the content item and the topics determined based on one or more characteristics describing frequency of presentation of each of the identified key terms in the content item and describing differences in presentation of each of the identified key terms by the content item relative to presentation of other terms in each content item;

generating one or more clusters based at least in part on the generated vectors, each cluster including one or more content items;

determining characteristic vectors for each cluster based at least in part on the generated vectors, a characteristic vector for a cluster based at least in part on one or more of the generated vectors included in the cluster;

retrieving a plurality of candidate content items each having the identified specific type of content item; and

selecting one or more candidate content items for presentation to the user based at least in part on a measure of similarity between each candidate content item and one or more of the characteristic vectors.

10. The computer-implemented method of claim 9, wherein identifying the plurality of content items with which the user of the digital magazine server previously interacted comprises:

identifying a plurality of content items having at least a threshold number of connections to the user of the digital magazine server user via the digital magazine server.

11. The computer-implemented method of claim 9, wherein identifying the plurality of content items with which the user of the digital magazine server previously interacted comprises:

identifying a plurality of content items having at least a threshold number of connections to additional content items with which the user of the digital magazine server previously interacted.

12. The computer-implemented method of claim 9, wherein generating the vector for each content item based at least in part on one or more topics determined from key terms included in a content item comprises:

associating a weight with one or more dimensions of the vector based at least in part on a type of the content item for which the vector was generated.

13. The computer-implemented method of claim 9, wherein determining characteristic vectors for each cluster based at least in part on the generated vectors comprises:

determining values associated with one or more dimensions of each vector generated for content items in a cluster;

determining weights associated with one or more dimensions of each vector generated for content items in a cluster; and

generating a characteristic vector for the cluster based at least in part on the determined values and the determined weights.

14. The computer-implemented method of claim 9, wherein selecting one or more candidate content items for presentation to the user based at least in part on the measure of similarity between each candidate content item and one or more of the characteristic vectors comprises:

generating vectors for each of the candidate content items, a vector generated for a candidate content item based at least in part on one or more topics associated with the candidate content item and having one or more dimensions with a value of a dimension based at least in part on a number of times a topic occurs in the candidate content item;

determining a measure of similarity between the vectors generated for each candidate content item and one or more characteristic vectors based at least in part on values of one or more dimensions of the vector generated for each candidate content item and values of one or more dimensions of the characteristic vectors;

generating a ranking of the candidate content items based at least in part on the determined measures of similarity; and

selecting one or more candidate content items based at least in part on the ranking.

15. The computer-implemented method of claim 14, wherein selecting one or more candidate content items based at least in part on the ranking comprises:

selecting one or more content items having at least a threshold position in the ranking.

16. The computer-implemented method of claim 9, selecting one or more candidate content items for presentation to the user based at least in part on the measure of similarity between each candidate content item and one or more of the characteristic vectors comprises:

selecting one or more candidate content items having at least a threshold measure of similarity between a candidate content item and at least one characteristic vector.

17. A computer-implemented method for recommending content items to a user of a digital magazine server, the method comprising:

identifying a specific type of content item for recommending to the user of the digital magazine server;

determining a time window corresponding to the identified specific type of content item, the duration of the time window determined based on a number of content items having the specific type included in a content store of the digital magazine server, different types of content item corresponding to time windows having different durations;

identifying a plurality of content items with which the user of the digital magazine server previously interacted during the determined time window corresponding to the identified specific type of content item;

identifying key terms from each of the plurality of content items;

generating one or more topics associated with each content item based at least in part on the identified key terms and characteristics describing frequency of presentation of each of the identified key terms in each content item and describing differences in presentation of each of the identified key terms by each content item relative to presentation of other terms in each content item;

generating a representation of each content item based at least in part on one or more topics associated with a content item, the representation of the content item having one or more dimensions with a value of a dimension based at least in part on a number of times a topic occurs in the content item;

generating one or more clusters each including one or more content items based at least in part on the generated representations;

determining characteristic representations for each cluster based at least in part on the generated representation, a characteristic representation for a cluster based at least in part on one or more of the generated representations included in the cluster;

retrieving a candidate content item having the specific type of content item;

determining a measure of similarity between the candidate content item and one or more of the characteristic representations; and

selecting the candidate content item for recommendation to the user based at least in part on the determined measures of similarity.