微博关注兴趣模型区间约简限定的社区检测

Microblog Community Detection Based on Interest Model with Limit Interval Reduction

为解决微博社区发现过程中的检测效果不佳和效率偏低的问题，提出微博关注兴趣模型区间约简限定的社区检测算法。首先，通过参数权值有限约简进行曲线分析，并利用凸优化方法进行优化，对参数最优值区间进行划分：其次，对参数最优值分区进行断点设计和顺序搜索，对参数取值范围进一步进行限定，所有参数均对应唯一约简权值，并可实现并行化多个分区同步参数优化，进而有利于多个分区信息均衡融合：最后，基于新浪微博社区发现数据，通过实验对比验证算法在微博用户社区检测的微博社区发现数量、内聚兴趣均值指标上的优势。

In order to solve the problem of detection effect and low efficiency in the process of micro blog community discovery, this algorithm is proposed based on micro blog attention interest model. Firstly, through the parameters of the weight of the finite reduction curve analysis, and the use of convex optimization method to optimize the parameters of the optimal value of interval division; Secondly, the parameter optimal value partition is designed and the sequential search is carded out, and the range of parameters is further defined. All parameters correspond to the unique value and can be implemented to optimize the parallel multiple partition synchronization parameters; Finally, based on the Sina microblogging community found data, by comparing the experimental results, the advantages of the algorithm are compared with the number of micro-blog community and the average index of interest in the micro-blog community.