Cloud manufacturing provides a new concept and model for manufacturing informatization. However, it still faces many challenges, especially in the matching of service resources and service requirements in reconfigurable production line. In this article, we propose a service description language based on extended OWL-S (Ontology Web Language for Services) to encapsulate geographically dispersed and formally heterogeneous manufacturing resources into cloud services. We also propose a Semantic-based service matching approach using BTM (Biterm Topic Model Technology) and Gibbs sampling algorithm. This approach performs service screening through topic similarity matching and parameter array similarity matching, and recommends the most suitable service set for cloud service users, providing a new method for resource matching in the reconfigurable production line resources. Finally, through an example of service encapsulation and matching, we verify the feasibility and efficiency of the proposed service encapsulation method and service matching algorithm.