Yuxi Ma

Context: Automated test generation remains challenging for programs with deeply nested execution paths, boundary-sensitive numerical constraints, and limited transparency in the generation process.Objective: This paper proposes MAST (Multi-Path Analysis and Solver-Guided Test Generation), a hybrid framework that combines large language model (LLM)-guided multi-path analysis with solver-guided synthesis to improve coverage-oriented automated test generation while making the generation process more interpretable.Methods: In a single LLM generation round, MAST produces multiple structured candidate-path hypotheses, each containing a path rationale, concrete inputs, and explicit constraints. Supported numerical constraints are translated into Z3 formulas to synthesize or refine concrete test inputs. The resulting inputs are then validated through isolated execution and structural coverage measurement. MAST is evaluated on seven datasets against four representative search-based, LLM-only, and hybrid baselines, with ablation experiments assessing the contributions of its core components.Results: MAST achieves the highest branch and line coverage across all seven evaluation datasets. On the Pypara project, it reaches 82.14% branch coverage and 85.71% line coverage, exceeding the best-performing baseline by 17.35 and 9.85 percentage points, respectively. Ablation results further show that removing solver-guided synthesis reduces line coverage on MBPP_HumanEval(S) from 85.48% to 79.38%, a decrease of 6.10 percentage points, supporting the contribution of constraint refinement to coverage-oriented test generation.Conclusion: By integrating structured semantic path reasoning, formal constraint refinement, and execution-based validation, MAST connects high-level code reasoning with low-level constraint satisfaction and provides an interpretable approach to automated testing for programs with complex control flow and boundary-sensitive conditions.

Federated domain generalization aims to learn a global model that generalizes well to unseen domains. Feature alignment with prototypes enhances domain-invariant representation learning, aligning inconsistent feature spaces caused by domain skew. However, we identify two key problems that limits feature alignment: local prototypes become unreliable due to mismatched batch normalization statistics and insufficient inter-class separation, while conventional prototype aggregation biases the global feature space toward overrepresented domains. Building on these findings, we propose FedCoda, a cross-domain feature alignment framework that calibrates local prototype geometry and debiases global prototype aggregation. FedCoda first performs calibrated feature alignment by transitioning to global batch normalization statistics in the later training stage, while introducing a prototype calibration loss that encourages well-separated class prototypes. It then applies fair prototype aggregation, which optimizes global prototypes by minimizing the variance of their distances to local prototypes, reducing dominance from domains with more clients. Extensive experiments show that FedCoda consistently improves generalization ability and representation learning under various settings, demonstrating the importance of jointly addressing prototype reliability and aggregation fairness for federated domain generalization.

The continuous evolution of information technology and network expansion pose significant challenges to traditional network operation and maintenance (O\&M), where fragmented, multimodal knowledge and isolated knowledge silos hinder intelligent automation. From the artificial intelligence perspective, we propose a novel entity alignment diagram construction method that semantically associates visual entities to enhance knowledge fusion and retrieval efficiency. This method is integrated with retrieval-augmented generation (RAG) and large language models (LLMs) to build an intelligent question-answering system tailored for network operations. From the engineering application perspective, we develop two specialized datasets to evaluate RAG performance and multimodal question answering in real-world network O\&M scenarios. Experimental results demonstrate that our method outperforms traditional multimodal RAG and pure LLM baselines in answer quality, effectively enabling knowledge fusion and vertical-domain multimodal alignment for practical network O\&M deployment.

To address issues commonly observed during the inference phase of large language models—such as inconsistent labels, formatting errors, or semantic deviations—a series of targeted strategies has been proposed. First, a relation label refinement strategy based on semantic similarity and syntactic structure has been designed to calibrate the model’s outputs, thereby improving the accuracy and consistency of label prediction. Second, to meet the contextual modeling needs of different types of instance bags, a multi-level contextual augmentation strategy has been constructed. For multi-sentence instance bags, a graph-based retrieval enhancement mechanism is introduced, which integrates intra-bag entity co-occurrence networks with document-level sentence association graphs to strengthen the model’s understanding of cross-sentence semantic relations. For single-sentence instance bags, a semantic expansion strategy based on term frequency-inverse document frequency is employed to retrieve similar sentences. This enriches the training context under the premise of semantic consistency, alleviating the problem of insufficient contextual information. Notably, the proposed multi-granularity framework captures semantic symmetry between entities and relations across different levels of context, which is crucial for accurate and balanced relation understanding. The proposed methodology offers practical advancements for semantic analysis applications, particularly in knowledge graph development.