Performance Analysis of Multilevel Indices for Service Repositories
There are many different index structures for service repositories, such as sequential index, inverted index and multilevel indices that includes three deployments. Different service sets maybe have different characteristics that may affect performance from different aspects. What characteristic could affect retrieval performance? How to select an optimal storage structure for a given service set? To address these issues, this paper analyses five indexing models and proposes expectation of traversed service count to estimate performance of service retrieval. The proposed expectation formulas of five indices reveal what different characteristics of a service set could affect its retrieval performance for different indices. Experimental results validate correctness of the proposed formulas.
Xu, Wei; Wu, Yan; and Liu, Lu. (2016). "Performance Analysis of Multilevel Indices for Service Repositories". 2016 12th International Conference on Semantics, Knowledge and Grids SKG 2016, 103-108. http://dx.doi.org/10.1109/SKG.2016.023