Nowadays, the vast majority of large organizations possess hundreds of different business processes (BPs), which are typically poorly documented. Furthermore, the relationships between the different types of processes are not clearly specified (Houy et al., 2011). Rebuge and Ferreira (2012) presented characteristics of processes, including the following: the dynamic and changing nature, complexity, interdisciplinary nature of the processes; interactions between the processes in different departments; and the requirement for obtaining experience, knowledge, and expertise for implementing the processes. Moreover, they stated that the traditional analysis of processes was time consuming. In addition, creating a common understanding of the processes between employees was difficult. The other problems related to this research work are presented in Table 1. With respect to the above-mentioned problems, data mining (DM) techniques can extract valuable patterns hidden in the high volume of BPs for recommending PI suggestions. In this regard, a process ontology concept can be considered to share the patterns extracted from the application of DM in PI to gain process ontology benefits. Pivk et al. (2014) stated that applying ontology for implementing DM in BPs has benefits: (1) sharing a common understanding between people, (2) re-using the domain knowledge, (3) making explicit domain assumptions, (4) differentiating between domain and operational knowledge, and (5) analyzing the domain knowledge. This paper contributes an integrated three-part, five-stage framework for applying the DM approach for PI under a process ontology concept. Figure 1 exhibits the conceptual model of the proposed framework including three parts as follows: 1. process ontology, 2. DM, and 3. PI. Each part has a series of attributes and behaviors. The attributes explain the characters of these parts and describe what they are. The behaviors describe the activities that these parts can implement.