查看文章

Many publications deal with profile matching across online social networks and the approaches become increasingly complex. Almost all of them rely on common profile attributes like names and hobbies or structural attributes like relations to other user profiles. These approaches require high effort concerning computation, because each profile of one network has to be compared to all profiles of the other network. Complex approaches are not well suited to handle large datasets. Therefore, we present an approach to significantly reduce complexity by exploiting special properties of dataset IDs. We provide a proof of concept by an implementation of the use case of matching user profiles accross Twitter and Instagram. Additionally to the complexity problem of existing approaches, many profiles with similar attributes often lead to a restrictive trade-off between precision and recall of the matching strategy. Furthermore …