Data Lineage is an open research problem. This is particularly true in data integration systems, where information coming from different sources, potentially uncertain or even inconsistent with each other, is integrated. In this context, having the possibility to trace the lineage of certain data can help unraveling possible unexpected or questionable results.In this paper, we describe our preliminary work about this problem in the context of the MOMIS data fusion system. We discuss and compare the use of lineage and why-provenance for the data fusion operator used in the MOMIS system; in particular we evaluate how the computation of the why-provenance should be extended to deal with Resolution Functions used in our data fusion system.