An Intelligent Platform for Software Component Mining and Retrieval.

Area of Science:

• Robotics
• Software Engineering
• Computer Science

Background:

• Developing robotic, IoT, and sensor-based applications is complex and time-consuming.
• Effective utilization of software components is crucial for these applications.
• Existing development frameworks lack efficient code reuse mechanisms.

Purpose of the Study:

• To propose a platform that efficiently searches and recommends reusable code components for robotic applications.
• To streamline the development process by facilitating code component reuse.
• To enhance the accessibility and adaptability of programming frameworks for robotics.

Main Methods:

• Developed a platform employing a machine learning approach to train a schema for locating and ranking source code snippets.
• Implemented a user-friendly interface for developers to input queries (specifications) for code search.
• Utilized the trained schema to rank code snippets within the top k results.

Main Results:

• The platform effectively ranks source code snippets based on relevance to developer queries.
• The proposed approach demonstrates superior performance compared to existing baseline methods.
• Evaluation results confirm the platform's efficiency in recommending suitable code components.

Conclusions:

• The developed platform significantly facilitates the reuse of code components in robotic application development.
• The machine learning-based approach for schema training and ranking enhances development efficiency.
• A user survey validated the practical viability and effectiveness of the proposed methodology.