ABSTRACT In recent years, many studies have focused on unsupervised data‐to‐text generation methods. However, existing unsupervised methods still require a large amount of unlabeled sample training, leading to significant data collection overhead. We propose a low‐resource unsupervised method called CycleRUR. This method first converts various forms of structured data (such as tables, knowledge graph(KG) triples, and meaning representations(MR)) into unified KG triples to improve the model's ability to adapt to different structured data. Additionally, CycleRUR incorporates a retraining module and a contrastive learning module within a cycle training framework, enabling the model to learn and converge from a small amount of unpaired KG triples and reference text corpus, thereby improving the model's accuracy and convergence speed. We evaluated the model's performance on the WebNLG and E2E datasets. Using only 10% of unpaired training data, our method achieved the effects of fully supervised fine‐tuning. On the WebNLG dataset, it resulted in an 18.41% improvement in METEOR compared to supervised models. On the E2E dataset, it achieved improvements of 1.37% in METEOR and 4.97% in BLEU. Experiments also demonstrated that under unified linearization, CycleRUR exhibits good generalization capabilities.

With the development of communication technology and Internet of Things, Industrial Internet of Things (IIoT) is proposed in the automation industry for complex scenarios. Blockchain is applied in IIoT to solve data security and privacy issues related to centralized data storage and processing. However, there are inevitably performance issues with throughput constraints when blockchain manages large amounts of device data. This paper proposes a blockchain-supported performance optimization framework for IIoT systems using deep reinforcement learning (DRL) methods. We model the blockchain performance optimization problem as a Markov decision process that optimizes the blockchain’s throughput by dynamically adjusting the block size and interval through DRL while satisfying security constraints. We use the double deep Q-network (DDQN) to deal with the dynamic and complexity of optimization problems due to the heterogeneity of equipment and diversified requirements. We also alleviate the overestimation problem caused by DQN. Meanwhile, we study the impact of the number of network layers and different activation units on the performance optimization method in DDQN. Finally, we prove that our work is feasible and effective through the case study based on actual IIoT scenario datasets. Experimental results demonstrate that our proposed scheme enhances blockchain performance in IIoT systems. The detailed qualitative comparison with related work demonstrates the superiority and innovation of our work and proves that it improves the shortcomings of existing work.

Recent years, the successful application of blockchain in cryptocurrency has attracted a lot of attention, but it has also led to a rapid growth of illegal and criminal activities. Phishing scams have become the most serious type of crime in Ethereum. Some existing methods for phishing scams detection have limitations, such as high complexity, poor scalability, and high latency. In this article, we propose a novel framework named phishing detection on Ethereum via augmentation ego-graph based on graph neural network (PEAE-GNN). First, we obtain account labels and transaction records from authoritative websites and extract ego-graphs centered on labeled accounts. Then we propose a feature augmentation strategy based on structure features, transaction features and interaction intensity to augment the node features, so that these features of each ego-graph can be learned. Finally, we present a new graph-level representation, sorting the updated node features in descending order and then taking the mean value of the top n to obtain the graph representation, which can retain key information and reduce the introduction of noise. Extensive experimental results show that PEAE-GNN achieves the best performance on phishing detection tasks. At the same time, our framework has the advantages of lower complexity, better scalability, and higher efficiency, which detects phishing accounts at early stage.

Sentiment analysis has a wide range of promising applications in software engineering, and the development of deep learning has demonstrated that the uniform representation of different modalities can improve the model performance of sentiment analysis. However, in practical applications, multimodal sentiment analysis always faces unsatisfactory situations, especially when the modality has missing samples, most models may fail. For example, social dynamics of technicians in developer communities can face modality unavailability due to privacy settings. Several existing works based on deep learning and regularization methods have explored the modal missing problem, but these works cannot balance the cases of modal general missing (rate < 50%) and severe missing (rate ≥ 50%), and do not consider the resource consumption during model inference. Therefore, in this paper, we proposed a prototype augmented multimodal teacher-student network (PAMD) to address the above issues. Specifically, a multi-level and multi-origin distillation strategy is used to minimize the required resources and inference time, and prototype augmentation is used to guarantee the performance of the model when a modality is severely missing. Extensive experiments are conducted on different benchmark datasets to explore a network that balances performance and resource consumption. And It achieves good results in different modalities of missing cases.

The parking lot is one of the important components of the intelligent transportation system (ITS). The current parking lots mainly use instant parking, which has low parking efficiency, during peak hours, which leads to traffic congestion. To guarantee the stable operation of parking lots, we propose a blockchain-enabled parking reservation framework, called BPR. Traditional parking reservation systems may exist the condition of malicious reservations, and resulting in wasted parking spaces. Therefore, we design a reputation mechanism to manage the parking reservation behavior of vehicles and reduce the number of malicious nodes. In addition, to balance the performance of the blockchain at different times (especially during peak hours), we use deep learning (DL) to dynamically adjust the block size to make the blockchain run more efficiently and stably. We deploy the system in Hyperledger Fabric and conduct effectiveness experiments. The comprehensive evaluation results and analysis show that the proposed reputation mechanism can effectively curb malicious nodes from reserving parking spaces and reduce the waste of parking resources. And the block size will be dynamically adjusted to balance the performance of the blockchain at different periods, this method is also applicable to other blockchain performance-sensitive scenes. Finally, this paper is compared with related work to demonstrate the innovation and feasibility of this work from various aspects.

Summary Short text similarity plays an important role in natural language processing (NLP). It has been applied in many fields. Due to the lack of sufficient context in the short text, it is difficult to measure the similarity. The use of semantics similarity to calculate textual similarity has attracted the attention of academia and industry and achieved better results. In this survey, we have conducted a comprehensive and systematic analysis of semantic similarity. We first propose three categories of semantic similarity: corpus‐based, knowledge‐based, and deep learning (DL)‐based. We analyze the pros and cons of representative and novel algorithms in each category. Our analysis also includes the applications of these similarity measurement methods in other areas of NLP. We then evaluate state‐of‐the‐art DL methods on four common datasets, which proved that DL‐based can better solve the challenges of the short text similarity, such as sparsity and complexity. Especially, bidirectional encoder representations from transformer model can fully employ scarce information of short texts and semantic information and obtain higher accuracy and F1 value. We finally put forward some future directions.

Abstract Named Entity Recognition (NER) for cyber security aims to identify and classify cyber security terms from a large number of heterogeneous multisource cyber security texts. In the field of machine learning, deep neural networks automatically learn text features from a large number of datasets, but this data-driven method usually lacks the ability to deal with rare entities. Gasmi et al. proposed a deep learning method for named entity recognition in the field of cyber security, and achieved good results, reaching an F1 value of 82.8%. But it is difficult to accurately identify rare entities and complex words in the text.To cope with this challenge, this paper proposes a new model that combines data-driven deep learning methods with knowledge-driven dictionary methods to build dictionary features to assist in rare entity recognition. In addition, based on the data-driven deep learning model, an attention mechanism is adopted to enrich the local features of the text, better models the context, and improves the recognition effect of complex entities. Experimental results show that our method is better than the baseline model. Our model is more effective in identifying cyber security entities. The Precision, Recall and F1 value reached 90.19%, 86.60% and 88.36% respectively.

With the rapid development of Internet technology and the advent of the era of big data, more and more cyber security texts are provided on the Internet. These texts include not only security concepts, incidents, tools, guidelines, and policies, but also risk management approaches, best practices, assurances, technologies, and more. Through the integration of large-scale, heterogeneous, unstructured cyber security information, the identification and classification of cyber security entities can help handle cyber security issues. Due to the complexity and diversity of texts in the cyber security domain, it is difficult to identify security entities in the cyber security domain using the traditional named entity recognition (NER) methods. This paper describes various approaches and techniques for NER in this domain, including the rule-based approach, dictionary-based approach, and machine learning based approach, and discusses the problems faced by NER research in this domain, such as conjunction and disjunction, non-standardized naming convention, abbreviation, and massive nesting. Three future directions of NER in cyber security are proposed: (1) application of unsupervised or semi-supervised technology; (2) development of a more comprehensive cyber security ontology; (3) development of a more comprehensive deep learning model.

Abstract Traditional process modeling focuses on modeling activities for functional requirements. For dependability requirements, a knowledge‐based aspect‐oriented software process modeling approach is proposed. First, we extend the pattern and context to the knowledge graph triplet structure to describe dependability‐oriented knowledge patterns. By applying the patterns, dependability requirements can be organized into dependability‐related activities that are integrated into the software process. Then, aspect‐oriented modeling patterns based on Petri nets are introduced to support the integration of these dependability‐related activities and model dependability‐oriented software processes. Finally, the modeling performance and the subjective usability of the patterns are evaluated by 110 students with different degrees. The results indicate that these two indexes are on the positive track. Hence, the patterns may be the backbone of dependability‐oriented software process modeling.