Argumentation in Crime and Security Domain: Argument Mining and Natural Language Interfaces to Automated Reasoning

Intelligence analysis refers to “the application of individual and collective cognitive methods to evaluate, integrate, interpret information about situations and events to provide warning for potential threats or identify opportunities.” Analysts need (1) to gather evidence—i.e. information about situations and events—and (2) to generate sensible hypotheses—i.e. the result of sense-making activities over evidence— and (3) to effectively communicate them.

The aim of this PhD project is thus threefold:
• Support evidence gathering: review and assess existing techniques for data mining that can be applied to evidence gathering in order to reduce the burden on analysts;
• Support evidence evaluation: develop techniques to transform the results of data mining into sensible arguments in favour of hypotheses (argument mining) in order to support analysts in their sense-making activities over evidence;
• Support knowledge sharing: develop natural language generation techniques for efficiently sharing data and knowledge across individuals taking into considerations policies and privacy concerns.
Hypotheses inform strategies for preventing threats or coping with critical situations. To identify them, analysts must combine several approaches to assess evidence, establish what is credible, and understand what additional evidence may be required. This needs to be done in a timely manner, which poses significant challenges for individual analysts. This combined reasoning process is resistant to automation which is the reason why we propose to integrate statistical machine learning with other artificial intelligence technologies.

Although analysts are not the decision makers, we saw in recent, tragic events the importance of effective intelligence analysis to inform decision. Every step towards the aforementioned aims will result in more efficient crime (including terrorism) prevention and control. Indeed, by providing intelligence analyst with reliable, efficient methods for evidence evaluation and knowledge sharing, we can improve the process of intelligence by reducing human errors due to biases and miscommunication.