This study proposes to exploit recent advances in data mining techniques to examine the strategies published on the Internet by public authorities and implemented in Paris. We first conducted a qualitative review of strategic documents, released over 15 years. We then used a list of key terms concerning flood risk management to compute a semantic network and identify clusters of terms that tend to appear in similar contexts. The analysis reveals an evolution of how policy and decision-making processes are depicted on the Web. The proposed methodology can be reproduced in other cities facing climate risks.