mCAF: a multi-dimensional clustering algorithm for friends of social network services.

Area of Science:

• Social Network Analysis
• Data Mining
• Algorithm Development

Background:

• Rapid growth of social network services (SNS) has led to a large number of user connections.
• Managing and clustering large friend networks on SNS has become a significant challenge for users.
• Existing methods may lack the flexibility and efficiency required for modern social network management.

Purpose of the Study:

• To propose an automated algorithm, mCAF, for effective friend clustering in social network services.
• To develop novel methods for defining inter-friend distances using diverse measurement sets.
• To enhance user experience by reducing the time and effort needed for social network friend management.

Main Methods:

• Development of the mCAF (mutual Clustering Algorithm Framework) algorithm for automatic friend grouping.
• Implementation of distance metrics tailored to different user-defined measurement sets.
• Comparative analysis of mCAF against existing clustering algorithms like SCAN.

Main Results:

• The mCAF algorithm demonstrates superior performance in friend clustering tasks.
• Experimental results show a 35.8% improvement in similarity and an 84.9% improvement in F1 score compared to SCAN.
• The algorithm facilitates more flexible and convenient friend management through customizable privacy settings.

Conclusions:

• mCAF offers an effective solution for the challenge of managing large social network friend lists.
• The algorithm's ability to define custom distances and implement group-specific privacy settings enhances its utility.
• mCAF significantly outperforms existing methods, providing a more efficient and user-friendly approach to social network organization.