Last decade is a time of rapid evolution of the Internet. Various human’s life areas are migrating into the cyberspace, where people exist and act through virtual identities - personalization of themselves. Groups of virtual identities with their relations and context form cybersocieties. Due to the unique cyberspace's structure, virtual identities are characterized by relatively high anonymity level. This cause the phenomena of the virtual multi-identities, where a single physical person incarnates a few virtual identities. This entails both positive and negative effects, and presented article concerns a negative one, which is deceptive opinion spam, generated by multi-identities of a single person. This problem constantly increases as relaying on the opinions from the WEB becomes very common. This article presents the concept which combines elements from various domains, linked in order to solve this problem and detect virtual multi-identities hiding in social networks such as web forums, blogs or recommendation portals. The paper describes a general concept and the architecture of the implemented system. At the end, the evaluation of the solution is carried out, based on the examples from the recommendation portal and web forum.