多属性决策方法在电诈犯罪嫌疑人认定中的应用

Multi-attribute Decision-making Approach Applied into Identifi cation of Suspects Involved with Telecom Network Fraud

为协助警方在侦破电信网络诈骗案件时实现对所获证据的有效聚合,得到多名涉案人员的嫌疑程度排序,提出了一种基于语言毕达哥拉斯模糊集的多属性决策方法。首先,根据已有线索确定相关涉案人员及属性信息;其次,基于语言毕达哥拉斯模糊集以定性的语言术语表示模糊信息,对不确定的属性信息进行统一描述;最后,在语言毕达哥拉斯模糊环境下,利用基于WOWA算子的软似然函数这一多属性决策方法对各属性值进行有效聚合,得到相关涉案人员综合得分值的结果排序。结果表明,相较于定量的评估形式,定性的语言术语评估形式更适用于实际电诈犯罪案件;同时实验结果表明,相较于基于OWA算子的软似然函数多属性决策方法,本文所采用的WOWA算子能在一定程度上避免极端数据带来的影响,并且能更好地软化权重值,使得软似然值的变化更趋于平缓。本文所提多属性决策方法可协助警方关联与案件有关的属性信息,在弱化决策者主观因素的前提下缩小人工甄别范围以获得涉案人员的嫌疑程度结果排序,本文首次将该技术应用于电信网络诈骗案例中,对电诈犯罪嫌疑人认定具有实际意义。

Currently,fraud cases often occur with the form of intelligence,mainly appearing as those telecom network swindles.For such cases,police investigation usually renders the presumable yet not clear evidence about which the relevant evaluation is mostly rooted on the investigators’experience and knowledge.Obviously,this manipulation exists with certain human subjectivity capable of dwelling into some extent of fuzzinesss.Therefore,it is valuable to introduce the linguisticfuzzy set into the operation course of quantifying the involving evidential data so as to reduce the impact of human subjectivity on the obtained results.Here,the linguistic-fuzzy Pythagorean multi-attribute decision-making approach was tried to construct on the basis of WOWA-operator soft likelihood function,purposed to both effectively handle the essential weight of each attribute relating to those case-involved people described with equivocal language and get rid of the infl uence of extreme data and subjective decision-makers.First of all,the case-involved people and relating attribute information were determined according to the existing clues.Secondly,the qualitative linguistic terms were adopted to express the fuzzy information with linguistic Pythagorean fuzzy setting,having the uncertain attribute information uniformly described.Finally,the multiattribute decision-making approach was put into the linguistic Pythagorean fuzzy environment through WOWA-operator soft likelihood function so that the value of each attribute was effectively defi ned and aggregated,consequently having the ranked comprehensive scores delivered about those case-involved people.The results showed that the qualitative linguisticterm evaluation excelled to the quantitative one,more suitable for application into investigating actual telecom fraud cases.Meanwhile,the results also revealed that it was the WOWA-operator multi-attribute decision-making manner adopted in this assay not the OWA-operator one that was able to get rid of the infl uence of extreme data to some extent and better smoothen the weighted value,making the change of soft likelihood value more fl uent.The multi-attribute decision-making approach proposed here can assist the police to associate the case-related attribute information,reduce the screening scope about case and get ranking of suspects while weakening the subjectivity of decision-makers,hence demonstrating practical signifi cance for identifi cation of suspects committing telecom fraud crime.