Scientific Experience

I began my Ph.D. studies in 2009 at Blekinge Institute of Technology, in a project with the goal to prevent malicious software. Within this project, we investigated the identification of malicious software using reputation systems for software (using simulations), as well as the classification of software based on End User License Agreements (using predictive machine learning methods). Further, the classification of e-mail based on online social network data was investigated (using predictive machine learning methods). In 2012 I defended my Licentiate Thesis, Decision support for the estimation of the utility of software and e-mail. The main theme of the thesis was methods and decision support systems to help users decide on the utility of software and e-mail.

In 2012 I started working in a project in collaboration with local law enforcement in Blekinge. The goal of the project was to develop a method for structured data collection from residential burglaries and to develop a decision support system. The research that I was mostly involved in concerned how machine learning approaches can be used to detect links between crime cases, i.e. detecting series of crimes. The methods that were investigated were clustering and classification techniques, with the results evaluated and analyzed using appropriate statistical tests. The project initially involved local and regional law enforcement but grew to involve law enforcement on the national level and multiple regional areas, as well as the Swedish National Forensic Center. The developed decision support system was positively evaluated and used by law enforcement departments in several areas in Sweden for two years. The feedback from these organizations has been very positive, and good cooperation exists.

The project, in collaboration with law enforcement, has been very successful and has received attention in the media, by the government, and by several law enforcement organizations (including internationally). The interest for the project from Nordic criminology and law enforcement researchers, and Swedish and international law enforcement organizations have been high. Sweden’s former minister of justice expressed interest and visited BTH to learn more about the project. A number of presentations have been held within the project to several governmental agencies (e.g., Swedish Coast Guard) and other companies (e.g., insurance companies), where the research results have received positive feedback. Two follow up projects have been financed by the Regional Council of Blekinge (2015), and from Vinnova (2016). The collaboration with law enforcement during both continuation projects has been very positive. I

Late 2014 I defended my Ph.D. thesis, On Descriptive and Predictive Models for Serial Crime Analysis. The thesis’s main theme is intelligent decision support systems for crime series analysis, focusing on residential burglaries. The work is mainly in applied machine learning. The thesis deals with data structures, comparison of methods (predictive and descriptive), and evaluation of method results. A thread throughout my research career has been the use of machine learning to solve research problems. The methodology used is mainly experimental, in combination with appropriate approaches, e.g., statistical tests, in order to evaluate the results.

Since my Ph.D I have been involved in several projects, often in collaboration with industry (e.g. Ericsson and Telenor). More details are available in the CV.