The paper "Root Cause Analysis in Microservice Using Neural Granger Causal Discovery" by Cheng-Ming Lin, Ching Chang, Wei-Yao Wang, Kuang-Da Wang, and Wen-Chih Peng addresses the challenge of identifying root causes in microservice systems, which are increasingly complex and interconnected. Traditional methods like the PC-algorithm fail to capture the temporal dependencies and contextual information in time series data, leading to inaccurate root cause identification. To address this, the authors propose RUN (Root Cause Analysis using Neural Granger Causal Discovery), a novel approach that integrates neural Granger causal discovery with contrastive learning. RUN enhances the backbone encoder to incorporate contextual information from time series data and uses a time series forecasting model to conduct neural Granger causal discovery. It also employs Pagerank with a personalized vector to efficiently identify the top-k root causes. Extensive experiments on synthetic and real-world microservice datasets demonstrate that RUN outperforms state-of-the-art methods in root cause analysis. The paper includes a detailed methodology, experimental results, and a case study to validate the effectiveness of RUN in microservice applications. The authors conclude by discussing future work, focusing on improving the model's scalability for larger datasets.The paper "Root Cause Analysis in Microservice Using Neural Granger Causal Discovery" by Cheng-Ming Lin, Ching Chang, Wei-Yao Wang, Kuang-Da Wang, and Wen-Chih Peng addresses the challenge of identifying root causes in microservice systems, which are increasingly complex and interconnected. Traditional methods like the PC-algorithm fail to capture the temporal dependencies and contextual information in time series data, leading to inaccurate root cause identification. To address this, the authors propose RUN (Root Cause Analysis using Neural Granger Causal Discovery), a novel approach that integrates neural Granger causal discovery with contrastive learning. RUN enhances the backbone encoder to incorporate contextual information from time series data and uses a time series forecasting model to conduct neural Granger causal discovery. It also employs Pagerank with a personalized vector to efficiently identify the top-k root causes. Extensive experiments on synthetic and real-world microservice datasets demonstrate that RUN outperforms state-of-the-art methods in root cause analysis. The paper includes a detailed methodology, experimental results, and a case study to validate the effectiveness of RUN in microservice applications. The authors conclude by discussing future work, focusing on improving the model's scalability for larger datasets.