To address critical safety challenges in visionbased autonomous driving under foggy conditions, this paper proposes FOG-DETR, a trustworthy object detection framework that synergizes detection quality enhancement, system reliability assurance, and security-driven design. First, we design an Intensity-Guided Spatial Adaptive Attention Convolution (IGSAAC) mechanism to enhance feature quality by dynamically reorganizing spatial distributions, enabling precise separation of clear and fog-affected regions. Combined with a dual-path histogram self-attention strategy, our method achieves reliable feature representation through globallocal dynamic aggregation. Second, a Small Object-Enhanced Pyramid Module (SOEPM) is proposed to improve detection reliability for safety-critical small targets (e.g., pedestrians and vehicles) by integrating multi-scale features via OmniKernel fusion. The DySample module further ensures operational stability by eliminating upsampling artifacts while maintaining real-time efficiency. Experimental results on the RTTS(Rainy Traffic Testing Set) dataset demonstrate that the optimized FOGDETR achieves 2.8 % and 2.1 % improvements in mAP0.5 and mAP0.5:0.95 metrics respectively compared to the baseline RTDETR, with nearly equivalent parameter count. These advancements establish FOG-DETR as a quality-assured solution for vision systems in security-sensitive scenarios like autonomous driving and surveillance.

Named entity recognition is the most basic task in natural language processing, and its quality directly affects the performance of downstream tasks. Due to the scarcity of annotated corpus and diverse species in the medical domain, as well as the continuous emergence of neologisms, the application of general domain entity recognition methods in vertical fields face the problem of inaccurate recognition of professional vocabulary. Proximity relation as a strong prior information is of great significance for the annotation task of named entity recognition. In view of the insufficient of proximity modeling in current work, we propose a named entity recognition model based on proximity relationship enhancement (PEM). Firstly, BERT is introduced to obtain the semantic features of text sequences. At the same time, the embedding vectors corresponding to lexical labels and co-occurrence matrices are input into the graph attention network. This network uses an information gate mechanism to model the proximity dependencies and filter the redundant features. Furtherly, we utilize the multi-head cross attention mechanisms to align and fuse the proximity features, and introduce BiLSTM to extract the contextual semantics. Finally, the fused features are fed into CRF for decoding. In order to verify the effectiveness of the model in this paper, comparative ablation experiments are carried out on four commonly used medical entity datasets, such as CCKS2018. The experimental results further demonstrate that the PEM model outperforms other recognition models and can effectively improve the accuracy of entity recognition with good robustness.

Social network texts contain a great deal of sentiment information. Such information reflects the personal attitudes and emotional dispositions for particular topics or events. However, not much comprehensive semantic information and not enough text data are used by traditional text sentiment analysis models, consequently, there are shortcomings in the analysis results. To handle this problem, in this paper, we propose a text sentiment analysis model NBWAB based on BERT-WWM-ATT-BiLSTM text classification. Our optimal model is constructed as follows. BERT-WWM is first used to dynamically encode the character-level and sentence-level features, and then Bi-LSTM is used to capture deeper semantic features of texts. Finally, these results are fused with the relevant multi-dimensional features of texts by multi-head-attention feature fusion skill. To further improve the performance of text sentiment analysis, we employ the ChatGPT data augmentation method to extend training datasets. To show the efficiency of our model, we have conducted experiments on three Chinese datasets: SMP2020-EWECT, Waimai_10k, and Weibo_senti_100k. The accuracy and F1 value of the model on the SMP2020-EWECT dataset (usual) are 80.76% and 77.61%, respectively, the accuracy and F1 value on the Waimai_10k dataset are 92.29% and 91.34%, respectively, and the accuracy and F1 value on the Weibo_senti_100k dataset are 98.10% and 98.24%, respectively. The results show that our model has advantages over the existing models in that more semantic information and more text data are considered for text analysis.

The software version information affected by the CVEs (Common Vulnerabilities and Exposures) provided by the National Vulnerability Database (NVD) is not always accurate. This could seriously mislead the repair priority for software users, and greatly hinder the work of security researchers. Bao et al. improved the well-known Sliwerski-Zimmermann-Zeller (SZZ) algorithm for vulnerabilities (called V-SZZ) to precisely refine vulnerable software versions. But V-SZZ only focuses on those CVEs of which patches only have deleted lines.In this study, we target Java Open Source Software (OSS) by virtue of its pervasiveness and ubiquitousness. Due to Java’s object-oriented characteristic, a single security patch often involves modifications of multiple functions. Existing patch code similarity analysis does not consider patch existence from the point of view of an entire patch, which would generate too many false positives for Java CVEs. In this work, we address these limitations by introducing a two-stage approach named VERJava, to systematically assess vulnerable versions for a target vulnerability in Java OSS. Specifically, vulnerable versions are calculated respectively at a function level and an entire patch level, then the results are synthesized to decide the final vulnerable versions. For evaluation, we manually annotated the vulnerable versions of 167 real CVEs from seven popular Java open source projects. The result shows that VERJava achieves the precision of 90.7% on average, significantly outperforming the state-of-the-art work V-SZZ. Furthermore, our study reveals some interesting findings that have not yet been discussed.

Software architecture forms a bridge between requirements and code. In this paper, by defining port operations, we use port activities to describe component-based software architectures. We can get the following benefits: 1) The representation of an architecture with the proposed formulism is simpler comparing with those by other ADLs. 2) An architecture is a semigroup to the component operations: composing and nesting. This result may be used to check the consistence and adaptability of two architectures. 3) The port expressions can be easily mapped to Petri net, so that the port-based process can be checked through the analysis of the Petri nets.