论文部分内容阅读
如何通过话单数据来对目标人员的社会关系进行自动合理的分析,是公安工作中面临的重要问题。为解决此问题,提出一种基于神经网络模型的社会关系分析算法。通过评价参数与调整因子来计算社会关系权值。提出双向通话频率、双向通话时长、非正常时段通话与漫游通话4个评价参数来反应不同的社会关系特征,并采用前馈神经网络思维来对输出结果进行误差修正。实验结果表明,算法可以准确地筛选出分析目标在各种社会关系下的优先联系对象。对侦查与监控工作有一定的实际意义。
How to conduct an automatic and reasonable analysis on the social relations of the target personnel through the phone bill data is an important issue in public security work. To solve this problem, this paper proposes a social relationship analysis algorithm based on neural network model. Calculate social relation weights by evaluating parameters and adjustment factors. This paper proposed four evaluation parameters such as two-way calling frequency, two-way calling duration, non-normal calling duration and roaming call to reflect different social relationship characteristics, and using feedforward neural network to correct the output error. The experimental results show that the algorithm can accurately filter out the priority objects of the analysis target under various social relations. The investigation and monitoring work has some practical significance.