A novel and efficient index based web service discovery approach

•We propose in this paper a new index scheme that aims at reducing the retrieval time comparing to MLIM (The Muti Level Index Model) and the inverted index.•We introduce architecture, called Multi Level Sorted Index (MLSI). The latter is based on building sorted sub-trees of service input keys, thus providing deterministic paths to reach lists of services enjoying the same input parameters. The main goal of MLSI is to improve the overall performances of the service discovery.•To improve further the performances, we extend MLSI with additional levels to speed up the access to the right list of services according to the configuration of its output parameters.•This solution called, Multi Level Sorted Index oriented Output (MLSIO), can be implemented in huge repositories when the number of lists of service per leaf in MLSI overtakes a given threshold.•To evaluate the performances of our solutions, we report the results of many simulations that compare MLSI and MLSIO with MLIM and the inverted index when it comes to service discovery, service addition and structure size. Nowadays, with the expansion of the number of web services, the discovery and the selection of the latter are becoming complex tasks. To address this issue, different methods (indexing, clustering, database etc.) have been proposed in the literature so far. In this paper, we deal with index based methods to manage web service repositories. Indeed, using indexes perspective in web services discovery can help to promote the web service composition. Within this context, we propose a novel index method named Multi Level Sorted Index (MLSI) to improve the run time of the discovery process. The implementation and the simulation of our method have shown the efficiency of our approach in comparison with the inverted index and Multilevel Index Model (MLIM) in term of retrieval and addition times.