Network Data Mining Algorithm of Associated Users Based on Multi-Information Fusion

To explore how related users can optimize the network mining algorithm, author proposes a user algorithm based on fusion of attributes and relationships. This method recommends key technical problems solutions information represented by multi-information explores research associated data algorithms. Research has shown that is 65% higher than previous methods. AUMA-MRL good performance under different overlaps. Also, since node embedding obtained neighborhood sampling, for new nodes in network, quickly obtain embedding, as well similarity vector between rest therefore, newly added be mined, robustness enhanced. Compared with existing classical algorithms, recall rate proposed increased 17.5% average, which effectively mine network.