Identification of real-best-pages on online social networks

1. A method comprising:

by one or more computing devices of an online social network, accessing a first place-entity cluster of a redirection graph, wherein the first place-entity cluster comprises a plurality of place-entity nodes, and wherein the plurality of place-entity nodes comprises an initial canonical place-entity node for the cluster, each other place-entity node of the first place-entity cluster being connected to the initial canonical place-entity node by a redirection edge;

by the one or more computing devices, calculating, for each place-entity node in the first place-entity cluster of the redirection graph, a cluster-score for the place-entity node, wherein the cluster-score indicates a quality of the place-entity node with respect to the first place-entity cluster;

by the one or more computing devices, identifying one or more place-entity nodes having a cluster-score greater than a threshold cluster-score;

by the one or more computing devices, receiving a selection of one of the identified place-entity nodes as a replacement canonical place-entity node for the first place-entity cluster, wherein the replacement canonical place-entity node is different from the initial canonical place-entity node; and

by the one or more computing devices, in response to the selection of the replacement canonical place-entity node that is different from the initial canonical place-entity node, updating the first place-entity cluster by redefining the first place-entity cluster to:

add at least one additional place-entity node of the redirection graph, wherein each added place-entity node has a duplication-value with respect to the replacement canonical place-entity node that is greater than a threshold duplication-value; or

remove at least one place-entity node from the first place-entity cluster, wherein each removed place-entity node has a duplication-value with respect to the replacement canonical place-entity node that is less than or equal to the threshold duplication-value.

2. The method of claim 1 , wherein the first place-entity cluster is associated with a place identified by the online social network.

3. The method of claim 1 , wherein the initial canonical place-entity node is identified by a human operator.

4. The method of claim 2 , wherein the human operator selects one of a subset of place-entity nodes associated with the place-entity cluster.

5. The method of claim 1 , wherein calculating the cluster-score for the place-entity node comprises:

determining an initial cluster-score for the place-entity node based on a class of the place-entity node; and

refining the initial cluster-score based on a number of social signals associated with the place-entity node to calculate the cluster-score for the place-entity node.

6. The method of claim 5 , wherein the class of the place-entity node comprises one of:

an official page of the online social network;

a page associated with an external website related to the place-entity node; or

an unowned page.

7. The method of claim 5 , wherein the social signals comprise check-ins, likes, comments, views, or reviews of a place-entity associated with the place-entity node.

8. The method of claim 5 , wherein selection of the replacement canonical place-entity node comprises:

sending the identified place-entity nodes to a human operator for selection of the replacement canonical place-entity node.

9. The method of claim 1 , further comprising selecting the first place-entity cluster, wherein the first place-entity cluster is selected based at least in part on a number of social signals associated with the place-entity nodes of the first place-entity cluster.

10. The method of claim 1 , further comprising selecting the first place-entity cluster, wherein the first place-entity cluster is selected based at least in part on a viewer-entity-pair value for a page associated with a place-entity node of the first place-entity cluster.

11. The method of claim 1 , further comprising updating the redirection edges of the first place-entity cluster by:

removing, for each other place-entity node of the first place-entity cluster, the redirection edge connected to the initial canonical place-entity node; and

connecting each other place-entity node of the first place-entity cluster to the replacement canonical place-entity node by an updated redirection edge.

12. The method of claim 1 , further comprising determining a quality-metric for the first place-entity cluster based at least in part on a precision-value or a recall-value for the first place-entity cluster.

13. The method of claim 12 , wherein the precision-value is (N 0 −N R )/N 0 , wherein N 0 is a number of place-entity nodes initially included the first place-entity cluster, and N R is a number of place-entity nodes removed from the first place-entity cluster when it is updated.

14. The method of claim 12 , wherein the recall-value is (N 0 −N R )/(N 0 −N R +N A ), wherein N 0 is a number of place-entity nodes initially included the first place-entity cluster, N R is a number of place-entity nodes removed from the first place-entity cluster when it is updated, and N A is a number of place-entity nodes added to the first place-entity cluster when it is updated.

15. The method of claim 1 , further comprising determining a quality-metric for the first place-entity cluster based at least in part on whether the initial canonical place-entity node is different from the replacement canonical place-entity node.

16. The method of claim 1 , further comprising:

providing, to a client system of a user of the online social network, the one or more identified place-entity nodes having the cluster-score greater than the threshold cluster-score; and

receiving, from the client system of the user, the selection of the one of the identified place-entity nodes as the replacement canonical place-entity node.

17. The method of claim 1 , further comprising:

selecting the one of the identified place-entity nodes as the replacement canonical place-entity node based at least in part on a cluster-score of the selected place-entity node.

18. The method of claim 1 , wherein the identified one or more place-entity nodes having the cluster-score greater than the threshold cluster-score comprises the initial canonical place-entity node.

19. The method of claim 1 , wherein the cluster-score of a place-entity node indicates a quality of a class of the place-entity node.

20. The method of claim 19 , wherein the class of the place-entity node indicates a source or a type of the place-entity node.

21. One or more computer-readable non-transitory storage media embodying software that is operable when executed to:

access a first place-entity cluster of a redirection graph, wherein the first place-entity cluster comprises a plurality of place-entity nodes, and wherein the plurality of place-entity nodes comprises an initial canonical place-entity node for the cluster, each other place-entity node of the first place-entity cluster being connected to the initial canonical place-entity node by a redirection edge;

calculate, for each place-entity node in the first place-entity cluster of the redirection graph, a cluster-score for the place-entity node, wherein the cluster-score indicates a quality of the place-entity node with respect to the first place-entity cluster;

identify one or more place-entity nodes having a cluster-score greater than a threshold cluster-score;

receive a selection of one of the identified place-entity nodes as a replacement canonical place-entity node for the first place-entity cluster, wherein the replacement canonical place-entity node is different from the initial canonical place-entity node; and

in response to the selection of the replacement canonical place-entity node that is different from the initial canonical place-entity node, update the first place-entity cluster by redefining the first place-entity cluster to:

add at least one additional place-entity node of the redirection graph, wherein each added place-entity node has a duplication-value with respect to the replacement canonical place-entity node that is greater than a threshold duplication-value; or

remove at least one place-entity node from the first place-entity cluster, wherein each removed place-entity node has a duplication-value with respect to the replacement canonical place-entity node that is less than or equal to the threshold duplication-value.

22. A system comprising: one or more processors; and a memory coupled to the processors comprising instructions executable by the processors, the processors operable when executing the instructions to:

access a first place-entity cluster of a redirection graph, wherein the first place-entity cluster comprises a plurality of place-entity nodes, and wherein the plurality of place-entity nodes comprises an initial canonical place-entity node for the cluster, each other place-entity node of the first place-entity cluster being connected to the initial canonical place-entity node by a redirection edge;

calculate, for each place-entity node in the first place-entity cluster of the redirection graph, a cluster-score for the place-entity node, wherein the cluster-score indicates a quality of the place-entity node with respect to the first place-entity cluster;

identify one or more place-entity nodes having a cluster-score greater than a threshold cluster-score;

receive a selection of one of the identified place-entity nodes as a replacement canonical place-entity node for the first place-entity cluster, wherein the replacement canonical place-entity node is different from the initial canonical place-entity node; and

in response to the selection of the replacement canonical place-entity node that is different from the initial canonical place-entity node, update the first place-entity cluster by redefining the first place-entity cluster to:

add at least one additional place-entity node of the redirection graph, wherein each added place-entity node has a duplication-value with respect to the replacement canonical place-entity node that is greater than a threshold duplication-value; or

remove at least one place-entity node from the first place-entity cluster, wherein each removed place-entity node has a duplication-value with respect to the replacement canonical place-entity node that is less than or equal to the threshold duplication-value.