Abstract This work has as a phenomenon the graffiti that decorates the freight trains that circulate in a transnational railway network. Graffiti writers participate in a complex symbolic competition in which they seek to become known by marking the largest number of trains, with the best executed interventions and with the largest size, as well as registering and publishing them on social digital networks, particularly Instagram. The Getting Up as a dynamic of symbolic competition was documented by Castleman (1984) and has established itself as an object of study in studies related to New York graffiti and answers questions such as: What shared values are found among graffiti writers? And, what is the engine that reproduces the practice of announcing the presence of an individual? It is from the conceptual framework of Jenkins' participatory culture and communities of practice (Jenkins et al., 2009) that we approach the graffiti writers of the New York tradition on freight trains. The main question of this work is to answer how to objectify a network of individuals who share an identity transnationally, without face-to-face communication and who, through constant participation, the formation of seniority and peer evaluation, consolidate a community. This paper aims to expose the inclusion of two machine learning models in the calculation of the weight of nodes and edges when graphing a network whose data source is the Instagram posts of graffiti writers on freight trains in the North America region. The machine learning models allows listing attributes of both the content of a publication or the profile itself to add weight to the node, as well as the type of communication that exists between these points. For the image analysis, two convolutional neural networks are used: First, a pre-trained model with the ImageNet data set allows finding common elements that will help understand and describe the context that was documented. Second, a custom-trained object detection model distinguishes the types of graffiti with which freight trains are marked. To approach the captios, a classification model is used that allows hashtags or fragments to be grouped into four referential categories such as geographic, identity, self-representation and emotional texts. This method makes it possible to ensure that there is a geographically dispersed group that shares symbolic elements in their publications, that document their interventions on the physical plane and that interact on the socio-digital plane, forming a circulation network of applied meaning. It is also possible to measure, and group, the frequency of references used by writers, the types of graffiti, the characteristics of the interventions and how they are evaluated by peers.