Visual-ISAM: A Visualization Method for Software Failure Analysis and Evaluation based on Knowledge Graph Utilizing Improved SALKU Model

Abstract

Software faults constantly appear during software development and evolution. The information on various platforms for bug knowledge recording, such as Stack Overflow, is mostly stored in weak entity relational database missing linkable relationships, which results in negative impacts on knowledge reuse. To enrich the relationships between entities and construct a software fault knowledge graph, we improve the SALKU model by considering the direction of prediction results between a pair of knowledge units, and utilize it to predict the class of linkable knowledge units. Experiment results show the improved model increases the ratio of knowledge unit pairs with equivalent link prediction results from 90.2% to 100% based on the premise of ensuring precision, recall, and F1-score. Eventually, we visualize the data from Stack Overflow in the knowledge graph based on the extracted relationships.