An Effective Fuzzy Recommender System for Fund-raising Management

In the social economics field that deals with the nonprofit organizations (NPOs), the fund-raising is a crucial activity that requires the management of a great number of quantitative and qualitative information regarding Donors and Contacts (i.e., potential donors). This data is normally stored in a structured database (DB) by each NPO, and it is clear that their effective processing by data science methods significantly improves the performances of the fund-raising campaigns. For this reason, the use of rigorous mathematical methods and decision support systems (DSS) has been playing a very important role in this context. The process of fund-raising is very complex and in part different depending on the characteristics of each organization. However, a common important feature is the role of the Contacts, and therefore, the method for turning the Contacts into actual Donors contextualized in the so-called giving pyramid is crucial from a strategic point of view. Recently, a recommender system (RS) has been proposed to optimize the Contacts’ management, by computing the similarity of each Contact with respect to the Donors. In this contribution, we enhance and complete this model by considering both a large DB and two significant extensions of the model, obtaining in this way an effective and whole fuzzy RS. With respect to the DB, the availability of information is effectively exploited. As for the algorithm, a proper similarity measure is defined, based on the specificity of the context. Moreover, a complete estimation of the Contacts’ characteristics is taken into account, by considering not only the frequency but the averaged amount of the gift as well, in the context of a nonparametric approach. The experimental results show the effectiveness of the proposed system.