If you click save then the text will be stored in database.
Do you want to save the case?
Yes No
Dechefr is developed to aid law-enforcement and security professionals, public safety officers, corporate and private security/intelligence analysts, HR-specialist, psychologists, psychiatrists, social workers, and researchers with threat assessment of written communication.
As a digital research assistant, Dechefr can be used to analyze any kind of written material such as social media conversations, manifestos or weblogs. The basic concept of Dechefr is to conduct a structured threat assessment of a text to assure that all relevant indicators are assessed in a standardized and objective manner. For an accurate assessment, you will need a text of at least half a page (approximately 300 words).
When a text is uploaded (supported files: txt, pdf, & docx), Dechefr extracts scores for 36 indicators that are used to assess the risk of the author committing targeted violence. These indicators cover seven risk assessment domains: Emotionality, Social Index, Leakage, , Identification, Preoccupation, and Linguistic alignment. A detailed description of the assessment domains is provided below.
The 36 indicators are used to generate a General Risk Level. The General Risk Level is automatically generated and fixed, thus, cannot be modified by the user. The General Risk Level is provided in four levels:
A number of key indicators are used to generate a profile: Anger, Anxiety, Disgust, Grievance, Self orientation, Othering, Social connection, Group orientation, Weapons, Violence, Source of Influence, and, Military terminology. Generated profiles can be compared with profiles from the normal population, known violent lone offenders, or profiles saved in the user archive (dashboard).
Most indicators in Dechefr are latent constructs that cannot be directly observed. Latent constructs have no absolute values and are, thus, meaningful only in relative terms. It is therefore more informative (not to say necessary) to present the scores of a specific individual in relation to the population or a subpopulation (norm samples). Therefore, for all extracted indicators, Dechefr provides the normative standing of that indicator for a given text in the form of percentile score. The population (or normal group) is selected to represent the internet and comes from various sources. This includes blogs, extremist web forums, general web-forums and other forms of social media.
The indicators that are used to create the profile are divided into different factors where each factor contains a number of indicators: Emotionality (Anger, Anxiety, Disgust, and Grievance), Social index (Self orientation, Othering, Social connection, Group orientation), Leakage (Weapons, Violence), Identification (Source of Influence, Military terminology).
Emotions play a significant role in our everyday life and are important markers for predicting behavior. Also, emotions are prime indicators of the interaction between the individual's way of thinking and the surrounding world. Emotions have also been emphasized by various scholars in predictors of violent extremism. Positive emotion words (e.g., pleasure, joy) are used to describe positive events, while negative emotion words (anger, disgust) are used to describe negative events. A high degree of emotion words while describing an event also seems to correlate with a high degree of immersion in whatever event one is describing. These categories of emotions can be used to assess the emotionality in a text, and information can also be found in how the frequencies of positive and negative emotion words relate to each other. Normally, natural conversation contains almost twice as many positive than negative emotion words, while extremist texts have much higher relative degree of negative emotion words, mostly anger words.
The use of pronouns in natural language has been examined in various studies, and pronouns have been linked to different aspects of personality and emotion. Frequent use of third-person plural (they, them, etc.) in a group suggests that the group is defining itself to a large degree by the existence of an oppositional group. Frequent use of third-person plural is reliable indicators of negative identification with an out-group, which that can be considered as a precursor of intergroup conflict and sometimes extreme behavior. The use of third-person plural pronouns in the analysis of online groups such as American Nazis and animal rights groups has been proven to be the best predictor of extremism as rated by independent judges.
Leakage, the communication of intent to harm a specific target to a third party, usually implies a preoccupation with the target and may signal planning, or implementation of, for example, an attack. Research suggests that leakage commonly occurs in cases of targeted violence, ranging from school shootings to attacks on public figures. Leakage can be intentional or unintentional, and more or less specific with regards to the act. Studies on public figure attacks and assassinations have found a suggestive pattern of leakage, in which an attack has often been preceded by indirect, conditional, or direct threats. In different studies, the occurrence of pre-attack leakage ranges from 46% to 67%, and even higher for school shootings.
Identification is a warning behavior indicating a desire to be a "pseudo-commando", have a warrior mentality, closely associate with weapons or other military or law-enforcement paraphernalia, identification with previous attackers or assassins, or identifying oneself as an agent to advance a particular cause. Identification includes identification with radical action and identification with a role model. Violent lone offenders often tend to identify themselves as a kind of warrior, a person who is prone to use structured violence for a "higher cause". In these cases, the use of military terminology and a strong interest in weapons and military strategies can be observed. In Dechefr, we measure the use of words that are related to military terminology since it could imply that the author have a desire to be a pseudo-commando. It has also been noted among previous cases of lone offenders that in many cases, they have been influenced by and to some extent identify themselves with other radical thinker or leader (influencers). In Dechefr this is measured by including mentions of potential influencers.