PSO-BLSTM: A Novel Approach for Superior Web Service Classification
Volume 28 , Issue 1 , June 2026 , Pages 96-114
Authors
Hawbash Abas Nabi 1 ; Kamaran Hama Ali Faraj 1
1 Department of Computer, College of Science, University of Sulaimani, Sulaimani, Kurdistan Region, Iraq
10.17656/sujpas.1115
Keywords
Abstract
ABSTRACT
This paper presents an advanced web service classification approach for web service classification using the Particle Swarm Optimization-based Bidirectional Long Short-Term Memory model. Web services are classified under different application scenarios according to methods to be followed for best recommendations and the discovery of respective services. For instance, models such as LSTM, B-LSTM, and CNN have been put in place for this purpose, but most of the time, evidently, there is a slight way of dealing with the complexity one is often subject to with the data of dynamic web services. These limitations can be bypassed through the PSO-BLSTM model, which captures the tremendous temporal learning capability of B-LSTM and the optimization efficiency of PSO. Our model shows superiority over the traditional LSTM, B-LSTM, and CNN with Top 1 and Top 5 precisions besides Precision, Recall, and F-measure from the experimental results obtained. Meanwhile, the proposed PSO-BLSTM model outperforms the added reference models on the classification accuracies with more service categories such as e-commerce, photos, chat, and medical services. Higher values that track well with most of the convergence curves from the Top 1 and Top 5 accuracy indicate that the model is well-trained with stable performance. All these confirm the classification potentials of PSO-BLSTM to make a more precise and effective discovery system of web services.
Statistics
- Article view189
- Downloads10
- First online25 June 2026
- Published at25 June 2026
- RIS
- BibTeX
- EndNote
- Mendeley
- APA (7th edition)
- MLA (9th edition)
- Chicago
- Harvard
- IEEE
- Vancouver