Taking advantage of the growing number of data sets released by governments, American news organizations now often publish news applications, such as online databases or interactive maps. As a consequence, many residents of American cities can get constantly informed about each crime, environmental pollution or new commercial service that occur in their city or neighborhood. Although some researchers now study how such artifacts and data-driven techniques affect journalism, very little is known about how citizens actually make sense of such information. Yet, what are the various types of collectives emerging from the access to this structured and localized information? How a group of individuals actually emerge as a "public" attributing collective meanings to seemingly scattered single occurrences? We investigate an online initiative carried out by the Los Angeles Times based on the reporting and mapping of every homicide committed in Los Angeles county in the last four years. Each crime is systemically presented on a webpage which is then open to discussion. This operation, which has attracted around 30,000 comments since its beginning, seemed to us as a good opportunity to investigate how collective judgments are emerging around criminal occurrences. To perform a systematic and exhaustive analysis of this large textual corpus, we developed and made use of dedicated lexicometric and machine learning tools. Stressing the importance of ecology in the process of providing collective meanings to single occurrences, our claim is that emerging publics are strongly shaped by the characteristics of each neighborhood.