Network information methods devices and systems

Abstract

Methods and systems for predicting links in a network, such as a social network, are disclosed. The existing network structure can be used to optimize link prediction. The methods and systems can learn a distance metric and/or a degree preference function that are structure preserving to predict links for new/existing nodes based on node properties.

1 - 74 . (canceled) 
     
     
         75 . A method for generating proposed link recommendations for output to requesting processes running on one or more processor devices connected to at least one computer network server through a connecting computer network, comprising:
 storing, on a data store that is accessible by the at least one computer network server, profiles and links, each profile of the profiles being a data set containing characteristics of a respective one of a plurality of entities, each link of the links being a data set that corresponds to a relationship of a predefined type between one of the plurality of entities to linked one of the plurality of entities such that some of the plurality of entities are linked to respective first entities and not linked to second entities, whereby each link corresponds to a linked pair of entities, the totality of links defining an relationship network;   the at least one computer network server programmatically training a classifier based on distance metrics, each distance metric characterizing a respective on of the linked pairs, wherein the distance metric is responsive to links other than ones corresponding to the linked pair the classifier being such that at least a substantial extent of a totality of the links can be derived from the classifier responsively to the profiles without the information content of the links, whereby the trained classifier contains all the structural information of the extent of the relationship network; and   receiving a profile corresponding to a new entity and generating at least one link representing the new entity.   
     
     
         76 . The method of  claim 75 , wherein the number of the at least one link is read from a data store storing predetermined data. 
     
     
         77 . The method of  claim 75 , wherein the generating includes, using the classifier to estimate a structure of a new network that includes the new entity including predicting a number of the at least one link. 
     
     
         78 . The method of  claim 75 , wherein the entities are members of a social network. 
     
     
         79 . The method of  claim 78 , wherein the social network is generated by a dating web site. 
     
     
         80 . The method of  claim 75 , wherein the entities are students, the links represent friends, and the new entity is an incoming student for whom there are no links. 
     
     
         81 . A non-transitory computer-readable medium comprising instructions stored thereon that, when executed by a processor, cause the process to implement a method for generating proposed link recommendations for output to requesting processes running on one or more processor devices connected to at least one computer network server through a connecting computer network, the method comprising:
 storing, on a data store that is accessible by the at least one computer network server, profiles and links, each profile of the profiles being a data set containing characteristics of a respective one of a plurality of entities, each link of the links being a data set that corresponds to a relationship of a predefined type between one of the plurality of entities to linked one of the plurality of entities such that some of the plurality of entities are linked to respective first entities and not linked to second entities, whereby each link corresponds to a linked pair of entities, the totality of links defining an relationship network;   programmatically training, by the at least one computer network server, a classifier based on distance metrics, each distance metric characterizing a respective on of the linked pairs, wherein the distance metric is responsive to links other than ones corresponding to the linked pair the classifier being such that at least a substantial extent of a totality of the links can be derived from the classifier responsively to the profiles without the information content of the links, whereby the trained classifier contains all the structural information of the extent of the relationship network; and   receiving a profile corresponding to a new entity and generating at least one link representing the new entity.   
     
     
         82 . The non-transitory computer-readable medium of  claim 81 , wherein the number of the at least one link is read from a data store storing predetermined data. 
     
     
         83 . The non-transitory computer-readable medium of  claim 81 , wherein the generating includes, using the classifier to estimate a structure of a new network that includes the new entity including predicting a number of the at least one link. 
     
     
         84 . The non-transitory computer-readable medium of  claim 81 , wherein the entities are members of a social network. 
     
     
         85 . The non-transitory computer-readable medium of  claim 84 , wherein the social network is generated by a dating web site. 
     
     
         86 . The non-transitory computer-readable medium of  claim 81 , wherein the entities are students, the links represent friends, and the new entity is an incoming student for whom there are no links. 
     
     
         87 . A computer network server comprising:
 a processor; and   a data store;   wherein the processor is configured to:   store, on the data store, profiles and links, each profile of the profiles being a data set containing characteristics of a respective one of a plurality of entities, each link of the links being a data set that corresponds to a relationship of a predefined type between one of the plurality of entities to linked one of the plurality of entities such that some of the plurality of entities are linked to respective first entities and not linked to second entities, whereby each link corresponds to a linked pair of entities, the totality of links defining an relationship network;   programmatically train a classifier based on distance metrics, each distance metric characterizing a respective on of the linked pairs, wherein the distance metric is responsive to links other than ones corresponding to the linked pair the classifier being such that at least a substantial extent of a totality of the links can be derived from the classifier responsively to the profiles without the information content of the links, whereby the trained classifier contains all the structural information of the extent of the relationship network; and   receive a profile corresponding to a new entity; and   generate at least one link representing the new entity.   
     
     
         88 . The computer network server of  claim 87 , wherein the number of the at least one link is read from a data store storing predetermined data. 
     
     
         89 . The computer network server of  claim 87 , wherein the processor is further configured to use the classifier to estimate a structure of a new network that includes the new entity including predicting a number of the at least one link. 
     
     
         90 . The computer network server of  claim 87 , wherein the entities are members of a social network. 
     
     
         91 . The computer network server of  claim 90  wherein the social network is generated by a dating web site. 
     
     
         92 . The computer network server of  claim 87 , wherein the entities are students, the links represent friends, and the new entity is an incoming student for whom there are no links.