Abstract. The amount of data on the Web is increasing. Current exact and complete techniques for matching complex query pattern against graph-structured web data have limits. Considering web scale, exactness and completeness might have to be traded for responsiveness. We propose a new approach, allowing an affordable computation of an initial set of (possibly inexact) results, which can be incrementally refined as needed. It is based on approximate structure matching techniques, which leverage the notion of neighborhood overlap and structure index. For exact and complete result computation, evaluation results show that our incremental approach compares well with the state of the art. Moreover, approximative results can be computed in much lower response time, without compromising too much on precision.