It is often the case that analyses of human interactive activity are lost once an article is written about the results obtained. Although it is clear that corpora are gathered in order to answer particular research questions and that already collected corpora are often not adapted for answering new research questions, it is still interesting to reflect upon the capitalization and exploitation of analyses that have been carried out. For example, comparison of analyses, validation of analyses or alternatives modes of visualization could be possible. This article proposes a model of designation and extraction of parts of human interaction corpora using the anchor and link concepts that allow for experimenting on the reuse of analyses of human interactions.