loading page

A genetic algorithm inspired method for semantic web service discovery using entropy-based deep neural network clustering and singular value decomposition
  • Manish Kumar Mehrotra,
  • Suvendu Kanungo
Manish Kumar Mehrotra
Birla Institute of Technology

Corresponding Author:[email protected]

Author Profile
Suvendu Kanungo
Birla Institute of Technology
Author Profile

Abstract

Abtract: Web Services play a crucial role in the realm of e-business applications in view of the rapid growth of web servce technologies. Effectively discovering the most pertinent web services from an extensive collection is pivotal for seamless application execution. However, extracting the most pertinent web services is still a challenging task due to poor query results with lack of semantics, low precision and recall rates. While past studies have employed various methods for web service discovery, this paper incorporates a collaborative filtering approach by employing singular value decomposition (SVD) matrix factorization to capture the principal sematics hidden behind a user query and the description in services. The proposed work aims to enhance the retrieval of relevant outcomes for user queries by leveraging a genetic algorithm inspired collaborative SVD (G-CSVD) matrix factorization technique. This is complemented by an entropy-based deep neural network clustering process, culminating in efficient service discovery. Post-clustering, the knowledge base is integrated into user queries, utilizing the SPARQL endpoint for data retrieval and manipulation within the resource description framework (RDF) database. As a result, the proposed methodology yields pertinent outcomes from the RDF data store in response to user queries.This research contributes an in-depth analysis encompassing execution time, service discovery time, cluster accuracy, precision, recall, F1-score, search time, execution time, root mean square error (RMSE), and mean square error (MSE). Furthermore, a comparative assessment against conventional methods highlights the efficiency and effectiveness of the proposed approach.