Content quality evaluation and classification

Patent metadata

Field	Value
Publication number	US-10311525-B1
Application number	US-201815912479-A
Country	US
Kind code	B1
Filing date	Mar 5, 2018
Priority date	Dec 31, 2014
Publication date	Jun 4, 2019
Grant date	Jun 4, 2019

Abstract

A social networking system classifies content items according to their qualities for ranking and selection of content items to present to users within, for example, a newsfeed. Low-quality content items that are unlikely to be interesting or relevant to a user may be distinguished though they may appear to be popular among users in the social networking system. The social networking system identifies within the content items one or more features that are indicators of the quality of the content items. The social networking system can use one or more classifiers to evaluate the content items based on the features, and it can compute a quality metric indicating the quality of a content item based on the result obtained from the classifiers. The quality metric can be used in the ranking and selection of a set of content items to provide to the user.

First claim

What is claimed is:

1. A computer-implemented method, comprising:
receiving a request for a plurality of content items to be presented to a user of a social networking system, the user connected to a plurality of other users in the social networking system; and
for each of a set of candidate content items for the user:
determining one or more classifiers for one or more features of the candidate content item, the one or more classifiers describing a quality of the candidate content item; and
computing a quality metric for the candidate content item based on the one or more classifiers, the quality metric indicating the quality of the candidate content item relative to other candidate content items of the set and indicating a likelihood that the candidate content item is junk or spam.

2. The computer-implemented method of claim 1 , further comprising:
accessing the set of candidate content items for the user in response to the request, each of the candidate content items having an initial ranking for the user relative to other candidate content items that affects whether the candidate content item is selected for presentation to the user, and
adjusting the initial ranking of each of the candidate content items to account for the quality of the content items according to the quality metric for each of the candidate content items of the set.

3. The computer-implemented method of claim 2 , further comprising:
selecting from the set of candidate content items, the plurality of content items to present to the user in a newsfeed based on the adjusted initial rankings of the candidate content items.

4. The computer-implemented method of claim 1 , wherein the one or more classifiers comprise classifiers for the candidate content item based on an originality of the candidate content item, a gaming nature of the candidate content item, an engagement of other users with the candidate content item, a quality of a hyperlink in the candidate content item, or an owner of the candidate content item.

5. The computer-implemented method of claim 1 , wherein the one or more features of the candidate content item are identified by determining an originality of the candidate content item based on comparing a fingerprint of content of the candidate content item to fingerprints of content of previously posted content items in the social networking system, wherein a match with one of the fingerprints of content of previously posted content items indicates that the content of the candidate content item exists in the social networking system.

6. The computer-implemented method of claim 5 , wherein a classifier of the one or more classifiers is determined based on a number of instances of the content of the candidate content item found in the social networking system, and wherein the quality metric is applied in ranking the candidate content items to decrease a ranking of the candidate content item relative to other candidate content items.

7. The computer-implemented method of claim 1 , wherein one or more features of the candidate content item are identified by evaluating whether content of the candidate content item solicits interest from the user.

8. The computer-implemented method of claim 7 , wherein a classifier of the one or more classifiers is determined based on an explicitness with which the content item solicits interest from the user, and wherein the quality metric is applied in ranking the candidate content items to decrease credit assigned by interest in the candidate content.

9. The computer-implemented method of claim 1 , wherein one or more features of the candidate content item are identified based on one or more of: a quality of engagement of other users with the content item, a number of shares of or comments on the candidate content item, whether a message is included when the candidate content item is shared, an average length of the message included when the candidate content item is shared, an average length of comments on the candidate content item, and an affinity of users who interacted with the candidate content item for a user who posted the content candidate content item.

10. The computer-implemented method of claim 1 , wherein one or more features of the candidate content item are identified based on one or more of: a link quality of a hyperlink included in the candidate content item, a ratio of clicks on or selections of the hyperlink by users, indications of interest or likes of the hyperlink by users, and a ratio indicating a lower link quality.

11. A computer program product comprising a non-transitory computer-readable storage medium containing computer program code for:
receiving a request for a plurality of content items to be presented to a user of a social networking system, the user connected to a plurality of other users in the social networking system; and
for each of a set of candidate content items for the user:
identifying one or more features of the candidate content item, the one or more features related to a quality of the candidate content item,
determining one or more classifiers based on the one or more features identified, the one or more classifiers describing the quality of the candidate content item, and
computing a quality metric for the candidate content item based on the one or more classifiers, the quality metric indicating the quality of the candidate content item relative to other candidate content items of the set and indicating a likelihood that the candidate content item is junk or spam.

12. The computer program product of claim 11 , where the computer-readable storage medium further contains computer program code for:
accessing the set of candidate content items for the user in response to the request, each of the candidate content items having an initial ranking for the user relative to other candidate content items that affects whether the candidate content item is selected for presentation to the user, and
adjusting the initial ranking of each of the candidate content items to account for the quality of the content items according to the quality metric for each of the candidate content items of the set.

13. The computer program product of claim 12 , where the computer-readable storage medium further contains computer program code for:
selecting from the set of candidate content items the plurality of content items to present to the user in a newsfeed based on the adjusted initial rankings of the candidate content items.

14. The computer program product of claim 11 , wherein the one or more classifiers comprise classifiers for the candidate content item based on an originality of the candidate content item, a gaming nature of the candidate content item, an engagement of other users with the candidate content item, a quality of a hyperlink in the candidate content item, or an owner of the candidate content item.

15. The computer program product of claim 11 , wherein the one or more features of the candidate content item are identified by determining an originality of the candidate content item based on comparing a fingerprint of content of the candidate content item to fingerprints of content of previously posted content items in the social networking system, wherein a match with one of the fingerprints of content of previously posted content items indicates that the content of the candidate content item exists in the social networking system.

16. The computer program product of claim 15 , wherein a classifier of the one or more classifiers is determined based on a number of instances of the content of the candidate content item found in the social ne

Assignees

• Facebook Inc

Inventors

• Owens Erich James
• Vickrey David

Classifications

• G06Q10/40 Primary

  Business processes related to social networking or social networking services · CPC title

• G06Q30/02

  Marketing; Price estimation or determination; Fundraising · CPC title

• G06Q50/01 Primary

  Physics · mapped topic