As a generalist geographical journal, cybergeo has for a purpose, among with the participation to the construction of an Open Science, to reach a large disciplinary audience. We propose to explore its disciplinary context by combining citation network exploration with semantic analysis. A database of citation neighbors of cybergeo articles is constructed through an automated data collection resulting in around 2.10^6 articles with abstracts. The extraction of relevant n-grams among the whole corpus allows to construct a semantic network by co-occurrences analysis. The optimisation of its size and modularity on construction parameters yields consistent disciplines by community detection, confirming the very rich scientific environment of cybergeo, that extends for example from economic geography and political sciences to biogeography and physical geography. We finally unveil epistemological informations such as occurrence proximities or citation links between disciplines, by the computation of community and article-level interdisciplinarity measures at the first order (abstract content) and second order (citation content).