Process, as an important knowledge resource, must be effectively managed and improved. The main problems are the large number of processes, their specific features, and the complicated relationships between them, which all lead to the increase in complexity and create a high-dimensionality problem. Traditional process management systems are unable to manage and improve processes with a high volume of data. Data mining techniques, however, can be employed to identify valuable patterns. With the aid of these patterns, suggestions for process improvement can be presented. Further, process ontology can be applied to share the process patterns between people, facilitate the process understanding, and develop the reusability of the extracted patterns for process improvement.This study presents a combined three-part, five-stage framework of data mining, process improvement, and process ontology. To evaluate the applicability and effectiveness of the proposed framework, a real process dataset is applied. Two clustering and classification techniques are used to discover valuable patterns as the process ontology. The output of these two techniques can be considered as the recommendations for improving the processes. The proposed framework can be exploited to support process improvement methodologies in organizations.