Abstract: To address the problems in the traditional method of measuring microblogger’s influence, such as the unclear user relationship and unclear time characteristics, a modified HITS algorithm is proposed based on the differences in influence from the perspectives of network structure, interaction behavior, and time. It selects the dynamic interaction intensity between users as the edge weight, revises the central value through the sigmoid function, and integrates users’ authoritative value and the central value to quantify their influence. As a result, the algorithm has increased adaptability, the results are more reasonable. It then tested the algorithm using Sina microbloggers. Compared with the traditional method of calculation, the new does improve in terms of the DCG value, precision rate, recall rate and F-score evaluation index.

Key words: HITS algorithm, Microblog user, Influences, Dynamic interaction, Time slice