Knowledge engineering methods, such as DESIRE, KARL and CommonKADS, offered a wide range of modeling formalisms to describe dynamic behavior of reasoning process of a KBS. However, the problems with portability and interoperability and reusability of models arise when these knowledge engineering methodologies employ their own proprietary representation. Furthermore, these traditional knowledge engineering methods fail to support the share and reuse of distributed inference resources over the Web. To tackle these problems, we present an approach to representing reasoning processes using process modeling approaches at different abstract levels. Firstly, the UML activity diagram is employed to describe reasoning process models at a high-level. Thus, portability and interoperability and reusability of the models can be ensured due to the advantage of Platform-independent models of UML. To enable the share and reuse of reasoning components over the Web, UML-AD based models are then mapped into OWL-S process language by using XSLT rules.