Social machine is a rather new approach to deal with relevant problems in society, blending computational and social elements into software. It can be an extension of the Semantic Web, creating processes in which people do the creative work and the machine does the data administration. This article presents a proposal to apply this approach in violence and criminality domain, a relevant matter to Latin America and Caribbean – LAC — countries. It extends Social Machines by applying two published strategies to obtain semantics over social networks data. The development procedure was documented to provide a systematic procedure and an example application is presented to identify violence and criminality events. The resulting procedure validation was done by testing against developed formal models in the research area. Criminal activity data extraction results were also compared to official data, in order to identify similarities.