ABSTRACT Context Open‐source communities have become a crucial driver of technological innovation and developer growth. While numerous deep learning‐based recommender systems exist, they often fail to provide accurate recommendations for identifying suitable developers who meet the task requirements of project development due to project popularity imbalance and developer ability bias. Recent advances in retrieval‐augmented generation (RAG) have shown promise in enhancing data augmentation and personalization for recommender systems in scenarios with data sparsity. Objective This study aims to develop RGPRec, a RAG‐enhanced graph neural network (GNN) for personalized, debiased task recommendations in open‐source communities. The model targets task recommendation quality, fairness, and coverage capabilities for long‐tail developers. Methods RGPRec leverages RAG‐enhanced large language models (LLMs) to retrieve project‐related documents and enriched developer attributes to improve node representations through ego graph structural learning. It also recognizes the importance of long‐tail developers—less involved in projects but still contributing significantly—often overlooked by traditional recommender systems. Moreover, RGPRec reasonably evaluates each developer's ability through a multilabel assessment based on an ego GNN to calculate a debiased rating, generating more personalized and debiased task recommendations. Results Evaluation using real‐world data from two famous open‐source communities of SourceForge and GitHub, indicates that RGPRec significantly outperforms state‐of‐the‐art (SOTA) approaches in rating and ranking performances. In addition, ablation studies demonstrate the necessity of each component in the RGPRec model. Conclusion RGPRec effectively enhances task recommendation quality in open‐source communities. By integrating RAG and LLMs with GNNs, RGPRec achieves superior data augmentation and personalization compared to traditional approaches. Therefore, RGPRec can be a valuable tool for promoting inclusiveness and efficiency in collaborations within open‐source communities.

Microservice-based systems often suffer from reliability issues due to their intricate interactions and expanding scale. With the rapid growth of observability techniques, various methods have been proposed to achieve failure diagnosis, including root cause localization and failure type identification, by leveraging diverse monitoring data such as logs, metrics, or traces. However, traditional failure diagnosis methods that use single-modal data can hardly cover all failure scenarios due to the restricted information. Several failure diagnosis methods have been recently proposed to integrate multimodal data based on deep learning. These methods, however, tend to combine modalities indiscriminately and treat them equally in failure diagnosis, ignoring the relationship between specific modalities and different diagnostic tasks. This oversight hinders the effective utilization of the unique advantages offered by each modality. To address the limitation, we propose TVDiag , a multimodal failure diagnosis framework for locating culprit microservice instances and identifying their failure types (e.g., Net-packets Corruption) in microservice-based systems. TVDiag employs task-oriented learning to enhance the potential advantages of each modality and establishes cross-modal associations based on contrastive learning to extract view-invariant failure information. Furthermore, we develop a graph-level data augmentation strategy that randomly inactivates the observability of some normal microservice instances to mitigate the shortage of training data. Experimental results on four datasets show that TVDiag outperforms the state-of-the-art methods in multimodal failure diagnosis by at least 20.16% and 3.08% in terms of \(HR@1\) and F1-score, respectively.

Popularity bias in mobile APP recommendations skews results by over-prioritizing trending APPs, obscuring niche yet relevant alternatives, and conflating user interest with social conformity. To address this problem, we propose TDIC (short for Time-aware Disentanglement of Interest and Conformity), a novel causal framework tailored for personalized, debiased mobile APP suggestions. TDIC employs causal graph analysis to isolate user interest from conformity within interaction patterns, incorporates item quality to refine popularity judgments, and integrates temporal awareness to track the fluid nature of trends. TDIC disentangles authentic user intents from dynamic social influences, fostering unbiased and precise recommendations. Evaluated on two real-world datasets, MobileRec and Myket, TDIC demonstrates superior performance over state-of-the-art baselines, achieving gains of up to 17.24% in Recall@20 and 7.14% in NDCG@20 on MobileRec, and 2.55% in NDCG@50 on Myket. These outcomes highlight the crucial role of temporal dynamics and quality adjustments in mitigating popularity bias for mobile APP recommendations. Our code is available at https://github.com/ssea-lab/TDIC.

The quality of edge computing microservices is significantly influenced by their ability to perceive the spatiotemporal dynamics of user locations. Traditional approaches to microservice deployment in edge environments often rely on manual adjustments based on user position and base station load, which introduces substantial complexity and inefficiency. To address these challenges, we propose a novel methodology for spatiotemporal-aware microservice pre-deployment utilizing large language models (SAMPLE). By leveraging the predictive capabilities of spatiotemporal large language models, our approach enhances the microservice’s spatiotemporal awareness through trajectory forecasting. Additionally, we introduce an automated framework for generating optimal microservice deployment strategies based on the spatiotemporal relationships between users and services. Experimental results demonstrate that the proposed method significantly improves service quality by autonomously sensing user movement and dynamically adjusting deployment strategies, enhancing both the efficiency and responsiveness of edge services. The implementation code and datasets are available at https://github.com/ssea-lab/SAMPLE.

With the rapid adoption of multimodal large language models (MLMs) in autonomous agents, cross-platform task execution capabilities in educational settings have garnered significant attention. However, existing benchmark frameworks still exhibit notable deficiencies in supporting cross-platform tasks in educational contexts, especially when dealing with school-specific software (such as XiaoYa Intelligent Assistant, HuaShi XiaZi, etc.), where the efficiency of agents often significantly decreases due to a lack of understanding of the structural specifics of these private-domain software. Additionally, current evaluation methods heavily rely on coarse-grained metrics like goal orientation or trajectory matching, making it challenging to capture the detailed execution and efficiency of agents in complex tasks. To address these issues, we propose KGCE (Knowledge-Augmented Dual-Graph Evaluator for Cross-Platform Educational Agent Benchmarking with Multimodal Language Models), a novel benchmarking platform that integrates knowledge base enhancement and a dual-graph evaluation framework. We first constructed a dataset comprising 104 education-related tasks, covering Windows, Android, and cross-platform collaborative tasks. KGCE introduces a dual-graph evaluation framework that decomposes tasks into multiple sub-goals and verifies their completion status, providing fine-grained evaluation metrics. To overcome the execution bottlenecks of existing agents in private-domain tasks, we developed an enhanced agent system incorporating a knowledge base specific to school-specific software. The code can be found at https://github.com/Kinginlife/KGCE.

Graphical User Interface (GUI) designs often result in fragmented elements, leading to inefficient and redundant code when automatically converted. This paper presents MMNet, a novel end-to-end model for grouping these fragmented elements, leveraging multimodal feature representations and advanced retention mechanisms to improve grouping accuracy. MMNet uses UI sequence prediction, enhanced by large multimodal models, and a multi-scale retention mechanism to build a UI encoder. This approach captures temporal dependencies and multi-scale features, improving multimodal representation learning. To address the lack of fragmented UI element datasets, we constructed a new dataset and enriched its visual information using advanced multi-modal large models. Given the complex nature of UI design prototypes, it remains challenging for models to effectively learn the relationships between different modalities. We have adopted a multi-scale retention mechanism to further refine the relationship modeling between UI elements. Evaluated on our dataset of 71,851 UI elements, MMNet outperformed three state-of-the-art deep learning methods, demonstrating its effectiveness and innovation. The open-source code and datasets are available at https://anonymous.4open.science/r/MMNet-2343.

AI programming has become a popular topic in recent years. Code suggestion, with code suggestion being a key capability of AI programming. Copilot, an “AI programmer” that provides code suggestions from natural language descriptions, has been launched by GitHub and OpenAI. By far, Copilot has been widely used by millions of developers. However, little work has systematically evaluated the correctness of Copilot’s suggestions. We conducted an empirical study on all 2,033 LeetCode problems to assess Copilot’s code generation across four mainstream languages: C, Java, JavaScript, and Python. We have found that: (1) 70.0% of problems received at least one correct suggestion, with language-specific rates of 29.7% (C), 57.7% (Java), 54.1% (JavaScript), and 41.0% (Python); (2) correctness decreases as problem difficulty increases, with acceptance rates of 89.3% (easy), 72.1% (medium), and 43.4% (hard); (3) acceptance rates vary across problem domains from 49.5% to 90.1%, while Graph problems challenge C and Python most, and Prefix Sum and Heap challenge Java and JavaScript most; (4) for the incorrect suggestions, we further summarize 17 types of error reasons accounting for their incorrectness and analyzed possible causes for why these errors occur. We believe our study can provide valuable insights into Copilot’s capabilities and limitations.

Mobile edge computing (MEC) is an emerging computing paradigm receiving growing attention. MEC significantly reduces latency by processing user requests on edge servers rather than cloud centers, making it ideal for real-time applications. However, due to resource limitations and user mobility, microservice requests may fail, especially when users move at high speeds. This paper introduces a new tool, MobileEdgeSim, to simulate microservice-oriented MEC environments. MobileEdgeSim integrates mobility prediction and service composition to enhance the pre-deployment of microservices. To evaluate MobileEdgeSim, we conducted a series of experiments comparing it to several state-of-the-art baseline approaches. We also conducted a user study to evaluate the tool’s effectiveness in real-world scenarios. Our results indicate that MobileEdgeSim significantly improves the success rate of both user requests and responses while reducing resource costs. MobileEdgeSim is available at https://github.com/ssea-lab/MobileEdgesim.

With the widespread adoption of 5G and Internet of Things (IoT) technologies, the low latency provided by edge computing has great importance for real-time processing. However, managing numerous simultaneous service requests poses a significant challenge to maintaining low latency. Current edge server task scheduling methods often fail to balance multiple optimization goals effectively. This paper introduces a novel task-scheduling approach based on Evolutionary Computing (EC) theory and heuristic algorithms. We model service requests as task sequences and evaluate various scheduling schemes during each evolutionary process using Large Language Models (LLMs) services. Experimental results show that our task-scheduling algorithm outperforms existing heuristic and traditional reinforcement learning methods. Additionally, we investigate the effects of different heuristic strategies and compare the evolutionary outcomes across various LLM services.

Code review is an essential component of software development, playing a vital role in ensuring a comprehensive check of code changes. However, the continuous influx of pull requests and the limited pool of available reviewer candidates pose a significant challenge to the review process, making the task of assigning suitable reviewers to each review request increasingly difficult. To tackle this issue, we present MIRRec, a novel code reviewer recommendation method that leverages a hypergraph with multiplex relationships. MIRRec encodes high-order correlations that go beyond traditional pairwise connections using degree-free hyperedges among pull requests and developers. This way, it can capture high-order implicit connectivity and identify potential reviewers. To validate the effectiveness of MIRRec, we conducted experiments using a dataset comprising 48,374 pull requests from ten popular open-source software projects hosted on GitHub. The experiment results demonstrate that MIRRec, especially without PR-Review Commenters relationship, outperforms existing state-of-the-art code reviewer recommendation methods in terms of ACC and MRR, highlighting its significance in improving the code review process.

Edge computing, a recently developed computing paradigm, seeks to extend cloud computing by providing users minimal latency. In a mobile edge computing (MEC) environment, edge servers are placed close to edge users to offer computing resources, and the coverage of adjacent edge servers may partially overlap. Because of the restricted resource and coverage of each edge server, edge user allocation (EUA), i.e., determining the optimal way to allocate users to different servers in the overlapping area, has emerged as a major challenge in edge computing. Despite the NP-hardness of obtaining an optimal solution, it is possible to evaluate the quality of a solution in a short amount of time with given metrics. Consequently, deep reinforcement learning (DRL) can be used to solve EUA by attempting numerous allocations and optimizing the allocation strategy depending on the rewards of those allocations. In this study, we propose the Dual-sequence Attention Model (DSAM) as the DRL agent, which encodes users using self-attention mechanisms and directly outputs the probability of matching between users and servers using an attention-based pointer mechanism, enabling the selection of the most suitable server for each user. Experimental results show that our method outperforms the baseline approaches in terms of allocated users, required servers, and resource utilization, and its running speed meets real-time requirements.

In edge computing, service providers aim to provide users with low latency, high reliability, and more secure services by deploying services to edge servers close to users. However, scheduling service requests in edge computing scenarios is challenging because the edge service nodes typically have limited resources. This limitation makes it difficult to simultaneously optimize multiple objectives, including resource utilization, total running time, and average waiting time. Existing methods are insufficient to provide intelligent scheduling services that consider multiple optimization objectives, resulting in unsatisfactory scheduling results. To address this problem, we propose PSNet, a deep reinforcement learning-based Pointer Scoring Network, for edge task scheduling. PSNet integrates multiple pointer networks to optimize multiple objectives and obtain final scheduling results. Experiments carried out on three publicly available real-world datasets show that our method can effectively solve the multiobjective optimization problem on edge service requests, with an 11% improvement over several state-of-the-art methods.

As an emerging computing paradigm, mobile edge computing (MEC) is receiving growing attention. In MEC, user requests on software applications are firstly sent to edge servers for processing, which can significantly reduce the latency compared with sending to cloud centers. Furthermore, a software application adopting the popular microservice architecture usually contains multiple intercommunicating microservices. This suggests that the software used by a moving user will invoke different microservices on different locations. However, a microservice request may fail as no corresponding microservice is deployed on nearby edge servers due to resource limitation and coverage limitation. Moreover, if the user is moving at high speed, the user may leave the coverage of the edge server before receiving a response. To address this issue, we propose a microservice pre-deployment approach by integrating mobility prediction and service composition. Our work aims to improve the success rate of both request and response for multi-users while reducing the resource cost of pre-deployment. Three groups of experiments demonstrate that our approach can significantly improve the performance of microservice pre-deployment compared with several baseline approaches.

Quality of Service (QoS) prediction is a hot topic in services computing, which has been extensively investigated in the past decade. Many approaches have been proposed to predict unknown QoS values of Web services according to their historical invocation records. These methods usually analyze each individual QoS data as a basic unit while ignoring the intrinsic characteristics of these time-series QoS data. In an extremely dynamic environment, how to capture the intrinsic and time-varying characteristics of QoS data from a finer-grained perspective becomes an essential issue to achieve accurate prediction. In this paper, we propose a hybrid QoS prediction approach by combining the Empirical Mode Decomposition (EMD) and the multivariate LSTM (Long Short-Term Memory) model. Our approach aims to capture the potential information in the historical sequence and perform accurate QoS forecasting. Experiments conducted on two realworld datasets show that our approach outperforms several state-of-the-art methods in QoS prediction performance.