A solution for discovering shared software components (C1-C9) of software products (PRODa-PRODe) installed on a data processing system is proposed. In order to achieve this result, for each software product the corresponding software components are aggregated into groups (Ga1-Ge1); each group includes software components that must be linked by a corresponding relationship (such as a co-location on the same operating system image). In this way, it is possible to assign a software component available on the system to its software product even when it may be used by more of them. Particularly, the operation is performed deterministically when the available software component is included in a single valid group (i.e., a group including all the required software components); in this case, the available software component (C2) is assigned to the software product (PRODb) associated with the single valid group (Gb1). Alternatively, when an (ambiguous) available software component (C4,C8,C9) is included in more valid groups, a score is calculated for each associated (eligible) software product according to the corresponding available software components; the eligible software product with the highest score (PRODc) is then selected for assigning the ambiguous software component.