Control Flow Graph Embedding Based on Multi-Instance Decomposition for Bug Localization
Authors: Xuan Huo, Ming Li, Zhi-Hua Zhou4223-4230
AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results on widely-used software projects demonstrate the effectiveness of our proposed CG-CNN model. |
| Researcher Affiliation | Academia | Xuan Huo, Ming Li, Zhi-Hua Zhou National Key Laboratory for Novel Software Technology, Nanjing University, Nanjing 210023, China {huox, lim, zhouzh}@lamda.nju.edu.cn |
| Pseudocode | No | The paper describes the model architecture and process but does not include formal pseudocode or an algorithm block. |
| Open Source Code | No | The paper does not provide an explicit statement or link for the open-source code of their methodology. |
| Open Datasets | Yes | The datasets used in the experiments are extracted from widely-used open source projects. All the bug reports and software code can be extracted from bug tracking system (Bugzilla) and version control system (Git), which have been widely used in previous studies (Zhou, Zhang, and Lo 2012; Saha et al. 2013; Ye, Bunescu, and Liu 2014; Huo, Li, and Zhou 2016). |
| Dataset Splits | Yes | In our experiments, 10-fold cross validation is repeated 10 times for each data set. |
| Hardware Specification | No | The paper does not provide any specific hardware details for running its experiments. |
| Software Dependencies | No | The paper mentions software components and methods like CNN, Deep Walk, LSTM, ReLU, Skip-Gram, SGD, and pre-trained word embeddings, but does not provide specific version numbers for any libraries or frameworks used (e.g., TensorFlow, PyTorch, Scikit-learn versions). |
| Experiment Setup | Yes | We then use traditional Re LU as activation function and the window size of convolutional filters is set as 3,4,5 and 100 feature maps is generated each filter. The number of nodes in LSTM is the same as the output of CNN. In addition, the drop-out method (Hinton et al. 2012; Krizhevsky, Sutskever, and Hinton 2012) is applied in the fuly-connected layers to prevent overfitting. All the parameters are learned by minimizing the loss function using stochastic gradient descent (SGD) based method. |