In this paper, the potentials of systematically linking philosophy of science with bibliometrics are investigated by exploring whether concepts developed within the structuralist theory of science can be used as interpretative basis for author co-citation studies. It is argued that clusters of co-cited authors cannot be interpreted straightforwardly as scientific communities nor as scientific generations. The first reason is that the respective constituents differ (authors vs. scientists), the second is that the co-citation relation generates non-Kuhnian communities, i.e. communities not sharing (...) a paradigm/theory-core. Instead, co-citation clusters can more adequately be taken to represent communities of common (epistemic) interest. Hence, the concepts of interest community and of discourse community are introduced into the structuralist framework making use of the notion of intended application. All in all, it becomes clear that a full-fledged theory of publication is the still missing prerequisite for further linking bibliometrics and philosophy of science at a large scale. (shrink)