The aim of the project RESOCIT is to analyse relational logics of scientific citations and references. It is a continuation, but also an alternative to recent works on social networks of scientific citations which analyse formal aspects of combination of intellectual sphere (citations) and social sphere (collaborations). Our project is innovative on two sides. Firstly, we consider citations as elements of relational dynamism and not only as indicators. Citations and references are an opportunity for scholars to build ties with others scholars or groups of scholars, to reinforce them, and sometimes to lighten them. Secondly, our methodology consists in elaborating a “mixed method” (quantitative and qualitative, synchronous and diachronous). This method is based on selecting a corpus of scientific publications and making interviews with their authors. Our objective is 1) to use scientific citations as name generators to understand the socio-cognitive network of the publication (intellectual, professional and social); 2) to complete this analyse with relational studies of publication process (writing, evaluation, cooperation, funding, links with previous and later publications); 3) to survey the diffusion of citations with particular attention to actors, supports and temporality of mediation chains that they draw. We will gathered 4 corpus of international scientific publications from the Web of Science of Thomson Reuters (N=150), from natural sciences (chemistry and biology) and social sciences (economy and sociology). In this project, citations (with all their social substance) are seen as an expression of scientific sociability and potential vector of transformation of scientific groups. The objective is to show that published texts determinate, consolidate or even undo relations between groups of scientists. Few works pay attention to relational circles of scholars and few of them use citations to understand these relations. When they do it, they tend to consider citations like indicators of visibility (impact factors) or like indicators of specialities (co-citations). The objective in this project is, to the contrary, to highlight the social and relational substance of citations practices. The purpose of this project is to develop a new mixed method of relational analyse of scientific citations which open new perspectives: better knowledge on scientific social network but also new tools for research evaluation. Indeed, because citations are more and more used as indicators for evaluating scientific activity, a qualitative and relational analyse is necessary for better understanding the bases of these practices.