In one aspect, the scoring operation for the intelligent application management system 430 of FIG. 4 may form the basis of a probabilistic calculation. The scoring model may initially provide a rules-based approach based user preferences or learned via a machine learning operation. The output of this scoring model may be extended into the creation of a trained probabilistic model with a feedback component for the purpose of reinforced weighting (e.g., machine learning). In one aspect, the initial scoring model for ranking each of the software applications according to a degree of importance, a degree of correlation, or a combination thereof in relation to the one or more the data sources 510, 520, and 530 to indicate either retaining the software application and/or decommissioning the software may be designed, developed, or started with following features, as illustrate in the follow format:

1) Account Name Vocabulary: V={w1, w2, . . . , wn}

2) Hostname Vocabulary: V={x1, x2, . . . , xn}

3) Component Vocabulary: V={y1, y2, . . . , yn}

4) Query: q=q1, . . . , qm where qi{V}

5) Document: di=di1, . . . , dimj where dij {V}

6) Document Type: V={z1, z2, . . . , zn}

6) Collection: C={d1, . . . dm}; and

7) Set of relevant documents: R(q){C}