Automatically Building Service-Based Systems With Function Relaxation.

Area of Science:

• Software Engineering
• Artificial Intelligence
• Database Systems

Background:

• Service-based systems (SBSs) are crucial in software engineering, with automated building methods using keywords being a focus.
• Existing keyword-based methods construct SBSs by extracting minimum-weight group Steiner trees from service repository graphs.
• A limitation of current methods is the inability to relax user function requirements, potentially leading to overly complex or incomplete systems.

Purpose of the Study:

• To address the limitations of existing keyword-based SBS building methods.
• To introduce and define the bounded skyline SBS building problem, allowing for relaxed function requirements.
• To develop adaptive and less restrictive automated methods for SBS construction.

Main Methods:

• The study defines the bounded skyline SBS building problem.
• Two novel algorithms are proposed, leveraging skyline query, dynamic programming, and lower bound pruning.
• Experiments utilize real-world datasets with keyword-labeled nodes to evaluate the algorithms.

Main Results:

• The proposed algorithms demonstrate time efficiency in automatically finding SBSs.
• The bounded skyline approach offers a more adaptive solution compared to traditional methods.
• Comprehensive experimental studies validate the effectiveness of the developed algorithms.

Conclusions:

• The bounded skyline SBS building problem offers a more flexible alternative to traditional methods.
• The developed algorithms efficiently construct adaptive service-based systems.
• Open-sourced datasets and code facilitate further research in automated SBS construction.