In the last years, we have witnessed vast increase of Linked Data datasets not only in the volume, but also in number of various domains and across different sectors. However, due to the nature and techniques used within Linked Data, it is non-trivial work for normal users to quickly understand what is within the datasets, and even for tech-users to efficiently exploit the datasets. In this paper, we propose a graph pattern based framework for realising a customisable data exploitation. Atomic graph patterns are identified as building blocks to construct facilities in various exploitation scenarios. In particular, we demonstrate how such graph patterns can facilitate quick understandings about RDF datasets as well as how they can be utilised to help data exploitation tasks like concept level browsing, query generation and data enrichment.