Recommending Analogical APIs via Knowledge Graph Embedding
Library migration, which re-implements the same software behavior by using a different library instead of using the current one, has been widely observed in software evolution. One essential part of library migration is to find an analogical API that could provide the same functionality as current ones. However, given the large number of libraries/APIs, manually finding an analogical API could be very time-consuming and error-prone. To date, researchers have proposed various automated analogical API recommendation techniques, among which documentation-based techniques have been intensively studied in the literature. However, existing documentation-based analogical API recommendation techniques have limited effectiveness by missing semantic similarity in the documentation and also suffer from the scalability issue.
In this work, we propose KGE4AM, a novel documentation-based approach that leverages knowledge graph (KG) embedding to recommend analogical APIs during library migration. In particular, KGE4AM proposes a novel unified API KG to comprehensively and structurally represent three types of knowledge in documentation, which could better capture the high-level semantics. In addition, KGE4AM then proposes to embed the unified API KG into vectors, enabling more effective and scalable similarity calculation. We implement KGE4AM as a fully automated technique with constructing a unified API KG of 35,773 Java libraries. We further evaluate KGE4AM in two API recommendation scenarios (i.e., with given target libraries and without given target libraries), and our results show that KGE4AM substantially outperforms state-of-the-art documentation-based techniques in both evaluation scenarios in terms of all metrics (e.g., 47.1%-143.0% and 11.7%-80.6% MRR improvements in each scenario). In addition, we further investigate the scalability of KGE4AM and find that KGE4AM can scale well with the increasing number of libraries.
Thu 7 DecDisplayed time zone: Pacific Time (US & Canada) change
14:00 - 15:30 | Models of Code and DocumentationResearch Papers / Journal First / Ideas, Visions and Reflections at Golden Gate C1 Chair(s): Gema Rodríguez-Pérez University of British Columbia (UBC) | ||
14:00 15mTalk | On the Usage of Continual Learning for Out-of-Distribution Generalization in Pre-trained Language Models of Code Research Papers Martin Weyssow DIRO, Université de Montréal, Xin Zhou Singapore Management University, Singapore, Kisub Kim School of Computing and Information Systems, Singapore Management University, David Lo School of Computing and Information Systems, Singapore Management University, Houari Sahraoui DIRO, Université de Montréal Pre-print Media Attached | ||
14:15 15mTalk | A Vision on Intentions in Software Engineering Ideas, Visions and Reflections Jacob Krüger Eindhoven University of Technology, Yi Li Nanyang Technological University, Chenguang Zhu Meta, Marsha Chechik University of Toronto, Thorsten Berger Ruhr University Bochum, Julia Rubin University of British Columbia, Canada Media Attached | ||
14:30 15mPaper | Automated Identification of Toxic Code Reviews Using ToxiCR Journal First Jaydeb Sarker Department of Computer Science, Wayne State University, Asif Kamal Turzo Wayne State University, Amiangshu Bosu Wayne State University, Ming Dong Wayne State University Link to publication DOI Pre-print Media Attached | ||
14:45 15mTalk | GrACE: Language Models Meet Code Edits Research Papers Priyanshu Gupta Microsoft, Avishree Khare Microsoft, Yasharth Bajpai Microsoft, Saikat Chakraborty Microsoft Research , Sumit Gulwani Microsoft, Aditya Kanade Microsoft Research India, Arjun Radhakrishna Microsoft, Gustavo Soares Microsoft, Ashish Tiwari Microsoft Media Attached | ||
15:00 15mTalk | Recommending Analogical APIs via Knowledge Graph Embedding Research Papers Mingwei Liu Fudan University, Yanjun Yang Fudan University, Yiling Lou Fudan University, Xin Peng Fudan University, Zhong Zhou Fudan University, Xueying Du Fudan University, Tianyong Yang Fudan University Pre-print Media Attached | ||
15:15 15mTalk | [Remote] CCT5: A Code-Change-Oriented Pre-Trained Model Research Papers Bo Lin National University of Defense Technology, Shangwen Wang National University of Defense Technology, Zhongxin Liu Zhejiang University, Yepang Liu Southern University of Science and Technology, Xin Xia Huawei Technologies, Xiaoguang Mao National University of Defense Technology DOI Pre-print Media Attached | ||