以BDI代理人架構為基礎之虛擬社群群體犯罪預防研究

Abstract

網路群體犯罪不同於組織犯罪般有結構的犯罪團體，亦非為了追求共同利益而合作的共犯夥伴，而是網路使用者自發性互動行為下逐漸浮現的群體近似犯罪行為。此現象普遍存在於當今各式各樣的網際網路社群，且以各種不同的樣貌與形式展現。本論文以差別接觸理論與社會學習理論為基礎，運用理論相關的元素與概念，以BDI代理人模式為方法來設計網路群體犯罪之模擬模式，透過動態模擬群體犯罪在不同條件下展現不同之面貌。更運用網路科學概念與分析方法，來分析犯罪關係網絡之特性，本研究藉由控制網路社群之使用者人數與初始犯罪率來觀察不同組合之下所演化的網路結構差異，並從四個衡量指標（犯罪技能平均數、群聚係數、前10%使用者平均連結度、連結度小於10之比率）來標示演化之網路結構的特徵，並觀察這些因素對網路虛擬社群群體犯罪的關連影響。研究結果預期將對虛擬社群群體經營者對於社群中群體犯罪之預防與有效管理將有所貢獻。

Collective crime is an emerging phenomenon along with collective intelligence soaring in recent years. In this paper, the notion of collective crime is regarded as a form of universally distributed crime originated from spontaneous interactions among online community users. The issues that collective crime addresses focus on deviant or criminal behavior existing in common groups or crowds rather than traditional topics at computer crime or cybercrime. In this research, we use the theories, "Differential Association" and "Social Learning", to form the conceptual foundation of the collective crime phenomena and model the dynamics of virtual communities with the BDI-based multi-agent approach to simulate the community social networks. Prevention of collective crime is then regarded as an evolving network function based on the micro level of function simulation and the macro level of network analysis with two controllable variables (initial crime rate and network size) and four indicators (average amount of crime skills, average cluster coefficient, average degree of top 10% users, and rate of users with degrees smaller than 10). Our research findings are believed to contribute to the prevention management of collective crime in virtual communities.