This paper presents a new method for extracting knowledge in pervasive information systems. We study the impact of the context of a user on the applications he runs on his smartphone. Our proposal relies on FCA and Galois lattices to manage the heterogeneity and complexity of contextual data. Manual analysis of Galois lattices in previous work has given promising results. We focus here on automating the process of interpretation of the Galois lattice, in order to generalize the knowledge extraction. We present original metrics illustrated on real data.