Home // International Journal On Advances in Software, volume 16, numbers 3 and 4, 2023 // View article

Lightweight Approach to Java Sample Code Recommendation System Using Apriori-Based Soft Clustering

Authors:
Yoshihisa Udagawa

Keywords: Recommendation System for Software Engineering; Maximal Frequent Itemset; Unsupervised Machine Learning; Soft clustering.

Abstract:
One effective way of learning programming techniques is to refer to sample code. However, it becomes difficult and time-consuming to find suitable sample code visually for a complex programming subject. To overcome this shortcoming, research and development of program recommendation systems have been actively conducted. This paper discusses a recommendation system for Java sample code using unsupervised machine learning techniques. The proposed system includes three major steps: (1) extracting invoked methods used in each sample code, (2) soft clustering the sample code by applying a data mining technique to the extracted methods, and (3) ranking the code in a cluster by calculating a weighted average concerning the extracted methods. Experiments using sample code related to a graphical user interface and string handling have confirmed the effectiveness of the proposed recommendation system. A generative artificial intelligence model can generate a description of the cluster using invoked method names that specify each soft cluster. The proposed recommendation system and a generative artificial intelligence model can collaborate for improving the programming education environment.

Pages: 254 to 266

Copyright: Copyright (c) to authors, 2023. Used with permission.

Publication date: December 30, 2023

Published in: journal

ISSN: 1942-2628