D-SCENT: Raising challenges to deception attempts using data scent trails

自9-11和7-7事件以来，恐怖主义一直是公众关注的主要问题。为了确保公共安全和保护英国经济，需要研究提供新的方法，在攻击被执行之前阻止它们，识别可能正在准备或正在进行攻击的人和网络，并提供明确的证据，可以用来证明审问、逮捕和起诉的合理性。在这项研究中，我们将调查是否可以从“气味痕迹”中识别和证明欺骗行为，也就是通过跟踪嫌疑人的行动、通信和行为来收集对嫌疑人一段时间活动的连贯描述。我们将开发软件，以推断哪些活动与嫌疑人的气味痕迹一致，哪些活动被排除。这些推论将使调查人员能够在实时(例如，鼓励嫌疑人放弃正在进行的攻击)和面谈期间(例如，指出嫌疑人的描述与气味痕迹证据之间的不一致，可能改变讯问过程)对嫌疑人进行质询。该项目将在进行欺骗性活动的人的背景下调查气味痕迹，以在对抗性的寻宝游戏中获得优势。这些游戏将与利益攸关方协商制定，以便为反恐背景提供一个不敏感的类比。玩家，通常是付费参与的本科生，将在游戏期间通过从支持地理空间定位设备的移动设备获得的位置和通信数据进行监控。将开发用于整合这些数据的新型软件，以建立玩家在玩游戏期间活动的气味痕迹。人工智能的方法将被结合起来，根据气味轨迹来推断在给定玩家的位置、轨迹、活动以及与其他人的联系的情况下，哪些活动是可能的，哪些不可能。我们设想由3个团队组成的游戏：A队代表对手，B队代表警察或普通公众，C队代表情报服务。A队通过在一定的时间内访问目标位置来得分，这是他们必须遵守的一套游戏规则，如果他们想赢的话。他们必须设法对B队隐藏违反规则的行为，B队通过成功阻止或识别A队的作弊而得分。C组可以通过向A组发送持有或可以向B组提供情报信息的气味踪迹的迹象来挑战A组。此外，气味踪迹的推论将支持C队在比赛期间关键时刻采访A队球员时，决定如何最好地证明或证伪怀疑。通过在比赛期间对球员进行观察，我们可以调查当人们面对揭露其欺骗行为的证据时，他们是如何改变行为的。我们还将在游戏期间的关键点采访玩家，模拟对嫌疑人的采访，从玩家对他们活动的描述中获取信息，然后根据他们自己的气味轨迹向他们提出挑战，这些线索与合法游戏的玩法一致或不一致。这将使采访和分析技术得到改进，并将为人们在面对欺骗后如何随后改变他们的行为提供线索。研究结果还将使我们能够测试来自法医心理学的假设，以确定如何最好地发现欺骗。这项研究还允许我们探索公众对反恐监测和监视的认识和反应。我们将通过一个由代表主要公共和学术机构的利益相关者和主题专家组成的咨询小组，确定与收集和使用人们通过公共空间流动的数据有关的伦理和法律问题。我们还将向游戏玩家和其他没有参与游戏的人进行问卷调查，以衡量在玩游戏和其他情况下对监测和监视的态度。

Since 9-11 and 7-7, terrorism has been a major public concern. To ensure public safety and to protect the UK economy, research is needed that offers new methods to foil attacks before they are executed, to identify people and networks who might be preparing for or undertaking an attack, and to provide clear evidence that can be used to justify questioning, arrests and prosecutions.In this study, we will investigate whether deception can be identified and proved from 'scent trails', that is, coherent accounts of suspects' activities over time compiled from tracking their movements, communications and behaviours. We will develop software to derive inferences about what activities are consistent with suspects' scent trails and what are ruled out. These inferences will allow investigators to challenge suspects, both in real time (e.g., to encourage suspects to abandon an ongoing attack) and during interviews (e.g., to point out inconsistencies between a suspect's account and scent trail evidence that might change the course of an interview).The project will investigate scent trails in the context of people undertaking deceptive activities to gain advantage in adversarial 'treasure hunt'-type games. The games will be developed in consultation with stakeholders to provide a non-sensitive analogy to counter-terrorism contexts. Players, typically undergraduate students paid for participation, will be monitored during games via positional and communication data obtained from mobile devices enabled with geospatial positioning devices. Novel software for integrating these data will be developed to build up scent trails of players' activities during game play. Methods of artificial intelligence will be combined to derive inferences from the scent trails about what kinds of activity are possible and impossible given a player's location, trajectory, activities and links with others.We envision games with 3 teams: Team A represent the adversary, Team B the police or general public, and Team C the intelligence services. Team A scores points by visiting target locations within a time limit under a set of game rules that they must violate if they are to win. They must try to hide rule violations from Team B, who score points by preventing or identifying Team A's deceptions successfully. Team C can challenge Team A by sending them indications of the scent trails that are held or can feed Team B intelligence information. Moreover, the inferences from scent trails will support Team C in deciding how best to prove or falsify a suspicion during an interview with Team A players at key points during the games.By conducting observation of players during games, we can investigate how people change their behaviours when they are confronted with evidence that reveals their deceptions. We will also interview players at key points during games as a simulation of interviews with suspects, eliciting from players accounts of their activities before presenting them with challenges based on their own scent trails that are either consistent or inconsistent with legal game playing. This will allow interview and analysis techniques to be improved and will provide clues as to how people subsequently change their behaviour after they have been confronted with their deception. The results will also allow us to test between hypotheses deriving from forensic psychology as to how best to detect deception.The research also allows us to explore public awareness of, and response to, monitoring and surveillance in counter-terrorism. With an advisory panel of stakeholders and subject specialists representing key public and academic bodies, we will identify ethical and legal issues associated with collecting and using data on peoples' movements through public spaces. We will also conduct questionnaire studies with game players and others not involved in the games, to measure attitudes to monitoring and surveillance in game-playing and other contexts.