Child sexual offenders on the internet: an analysis of online communication with real victims

Child sexual abuse is a burning social issue and has a major psychosocial impact on a child's future life. Global statistics show that this phenomenon is partially shifting to the cyberspace. These conditions enable offenders to have a wider reach and remain anonymous, thus increasing the risk of abuse. According to the international study EU Kids Online, 24 % of children in the Czech Republic have received unwanted sexual requests on the Internet, which is higher than the European average. Although the strategies and manipulative techniques used by offenders are often theoretically explored to improve the focus of preventive measures, the real online communications of offenders and victims have not been widely used so far, although they are more authentic and thus have better informative value. This study examines what the real online communication between the offender and the victim looks like and what its specifics are depending on the level of risk of the offender's behavior. The research sample includes 60 unique anonymized Facebook chat communications between individual offenders and victims (the average total number of messages per chat is 125). The data was acquired from a real documentary experiment that was filmed in the Czech Republic in 2018. In this experiment, three adult actresses acted as young girls on social networks and the document staff recorded how the offenders interacted with them. The behavior of the actresses was standardized according to the code of ethics (they never addressed the offenders first, they emphasized their pretended age in communication, their behavior was not seductive). The actresses were styled in the role of thirteen-year-old girls, and their communication was also adapted to that age. For the purposes of this study, the Facebook chat communications have been sorted and selected according to the following criteria: the offender sent at least 10 messages, the offender appears to be an adult, it was not a group chat. Chat communications were then categorized according to the risk level of the offender's behavior into low-risk and high-risk groups. The following criteria were used to assess the offender's risk (based on existing risk-assessment tools): use of sexually explicit words, use of manipulative strategies, planning a real encounter with the victim, sending sexually explicit photos or videos and questioning the victim's sexuality. Differences in communication styles of the low-risk and high-risk groups were further analyzed by the LIWC software, with emphasis on differences in the frequency of use of different word categories (affective processes, social processes, cognitive processes, biological processes, personal concerns etc.). The results are currently being analyzed and will be presented at the conference. The results may contribute to a better understanding of the child sexual abuse on the Internet based on data acquired from real online interactions of offenders and victims. This knowledge can be further used in the development of preventive tools that could, based on the principles of machine learning, detect risky communication on the Internet in real time. The authors declare that there is no conflict of interest.