Scalability of findability: effective and efficient IR operations in large information networks

Scalability of findability: effective and efficient IR operations in large information networks is a simulation paper about distributed retrieval with a large number of nodes (100, 1000, 10000) but with an incredibly small number of documents (4M?). Different strategies of search are compared including Random Walk, TFxIDF similarity, Degree, etc. All the experiments are interesting but I don't like the very small dataset which is un-realistic in any industrial context.