Identity recognition method and device of identical names in multiple networks

Abstract

The invention discloses an identity recognition method and device of identical names in multiple networks. The method comprises the following steps: obtaining user identity information and user identity corresponding relations in multiple networks, using a user identity information set with known user identity corresponding relations as a training set, establishing a lowest energy model based on user behavior similarity according to the user identity information in the training set, and obtaining energy factors and a matching relation classifier; matching two pieces of arbitrary user identity information according to the matching relation classifier, and solving the energy matrix to obtain a matching result of single prediction; integrating the matching results solved in multiple times to obtain the user identity corresponding relations. The method and the device disclosed by the invention can be used for confirming the identity of information having different user identity information while having the same names in the multiple networks to improve the accuracy of a statistical result, the adopted algorithm is efficient, the calculation process is fast, and with the increase of sample databases, the accuracy rate of a calculation result is continuously improved.

Description

technical field [0001] The invention relates to the technical field of information processing, in particular to a method and device for identifying the same name in multiple networks. Background technique [0002] Generally, the same user has different identity information registered in different networks, such as e-mail address, phone number and other information. For example, in the field of scientific research, there are often a large number of scientific and technological workers working together in multiple scientific research teams at the same time, which leads to the personal information used by the same person when publishing academic achievements, such as e-mail address, unit, address, etc., It may not be the same, that is, the same person name has different identity information. When summarizing relevant academic achievement information in the field, since it is difficult to judge whether these same names are the same person, such redundant information will direct...

AI Technical Summary

Problems solved by technology

[0003] Traditional social network analysis methods usually only consider the behavior characteristics of users in a single network (such as working in a university), ignoring the association that users may be in multiple networks, for example, a user can be active in universities, state-owned enterprises, and scientific research at the same time. Institutions and social research institutions, and each social network has different identities, social circles and research content, the behavior analysis method for a single network cannot be applied to this multi-layer network environment

Images