Integrating external knowledge with traditional spoken language understanding (SLU) models can effectively mitigate the ambiguity in user utterances in real-world scenarios. Knowledge graph, as a common source of external knowledge, encapsulates entities enriched with diverse attribute information. Nevertheless, existing models consider all entities as relevant, which introduces significant noise into the input. Additionally, not all attribute information of the entities is essential, resulting in considerable noise and redundancy. In this article, we propose a Noise-Removal of Knowledge-Enhanced (NRKE) framework for SLU, which involves two different types of denoising. The first approach involves hard denoising via entity selection, where we leverage a small clean dataset and introduce a BERT-based auxiliary model to filter out entities unrelated to user utterances, effectively eliminating noisy entities. In addition, we further refine entity selection by incorporating Large Language Models (LLMs) to assist in filtering out entities unrelated to user utterances. The second method involves soft denoising through the selection of entity attribute information. This approach utilizes a keywords-based local semantic selection that gives greater weight to relevant local semantics associated with specific keywords. This allows us to capture task-related information from the chosen entities, thereby minimizing noise and redundancy. To evaluate the generalization capability of existing knowledge-enhanced SLU models, we construct a new dataset named KGCAIS. The experimental results show that our NRKE achieves better performance than the competing models on both the PROSLU and KGCAIS datasets.

In the context of the rapid evolution of social media, zero-shot stance detection has become an essential task in Natural Language Processing, particularly in effectively identifying attitudes toward unseen targets in texts using large language models. However, existing methods have not fully taken advantage of Large Language Models (LLMs) capabilities in complex reasoning and contextual understanding, often leading to cumbersome approaches. This paper introduces a new framework named Refined-Attitude Reasoning (RAR), which combines example reasoning and attitude refinement techniques to significantly improve the performance of LLMs on stance detection tasks. Specifically, RAR first constructs a case library using information from the training dataset, retrieving examples similar to new input contexts by calculating a mixed weighted representation of text-target pairs. Second, it refines the expression of attitudes through backward reasoning from the cases and extracts broader attitude words to enhance the accuracy of stance identification in texts. Finally, during the generation phase, it combines optimized attitude words with examples for forward stance reasoning and implements a checking mechanism to ensure the accuracy and rationality of the generated results. Experiments show that our RAR framework achieves an overall F1 score exceeding that of other baseline methods on the zero-shot stance detection dataset VAST, demonstrating the importance of fine-grained attitude refinement and in-context learning in enhancing model performance.

Graph contrastive learning plays a crucial role in many fields, such as social network analysis and recommendation, and has become a hot research area in recent years. The key to improving the quality of contrastive learning is to obtain high-quality sample pairs that reflect the structural features of the data more effectively. Existing methods typically sample based on node similarity and can only select negative samples of a fixed difficulty level, leading to issues like false positives and false negatives. This study proposes a novel paradigm called Hardness-aware Negative Sampling Graph Contrastive Learning with LLMs. We outline the criteria that the adaptive hardness sampling method should follow and provide concrete instances. By adaptively selecting negative samples of appropriate hardness during the training process, this method effectively mitigates false negative and false positive problems. Additionally, we leverage large language models to integrate data with supplementary textual attributes. Extensive experiments and analyses demonstrate the superiority and effectiveness of our proposed method.

Localization security is crucial to the widespread applications of wireless sensor networks (WSNs) in various fields. This article mainly studies the issue of distributed localization in WSNs subject to deception attacks, in which the attacker randomly compromises communication channels and injects false data, resulting in the codification of data received by sensor nodes. A distributed iterative localization algorithm based on detection-holding strategy is proposed with the help of barycentric coordinate representations. This algorithm detects modified data in communication links through residual detection and communication encryption. It is proved theoretically that the proposed localization algorithm can achieve accurate convergence to the sensors' locations under general random false-data-injection attacks. Finally, the algorithm performance is demonstrated through simulation examples.

In the field of ancient Chinese text, extracting and analysing temporal and geographic information are crucial for understanding the personal experiences of historical figures, the development of historical events, and the overall historical background. Currently, named entity recognition(NER) strategies such as BERT+CRF are used to extract temporal and geographic information from ancient Chinese text. However, ancient Chinese text covers a vast time span, and the temporal and geographic entities constantly evolve and change, making it difficult to extract these entities from text. This paper proposes a temporal and geographic extraction model for ancient Chinese text, enhanced by time-series external knowledge base. The extraction of proprietary nouns and general structures are divided into two independent networks. An external database is applied to enhance extraction of proprietary nouns and reduce noise for general structure inference. We constructed address trees and chronological tables containing commonly used places and time-related keywords from different periods and collected 12,000 texts spanning 3,000 years for extensive training. Overall, our research highlights the importance of external knowledge base for ancient Chinese NER, and provides new ideas for research in related fields.

Nowadays, with political ideologies becoming increasingly polarized, detecting political perspectives has become more and more crucial. Previous studies mainly focused on reasoning with real-world entities as background knowledge, while they fail to effectively model hierarchical relationships in the text structure and leave out the document-level knowledge such as topics in a news article. To overcome these limitations, we propose KMGN, a novel Knowledge-enhanced Multi-Granularity Interaction Network for political perspective prediction which consists of two key components: (1) a knowledge infusion method for learning both the topic representations and political entities of a news article (2) a multi-granularity interaction network to learn semantic representations for words, sentences, and document and integrate the infused knowledge across different granularities. Extensive experiments shows that our proposed approach is effective in predicting the political stance in a news article. An ablation study demonstrates the contribution of each component in our model.

Few-shot event detection (ED) aims at identifying and typing event mentions from text with limited annotations. Most existing methods for few-shot ED use event ontology and related knowledge to construct prototypes and fail to fully leverage the rich knowledge of pre-trained language models (PLMs) which could help improve the representation of prototypes. Motivated by this, we propose an prompt-enhanced prototype framework which combines prototype and prompt for few-shot ED. Considering the scarcity of labeled data, we also introduce contrastive learning to enrich prototypes. Specifically, we use heuristic rules to align FrameNet with annotated data to get corresponding prompts for each event and convert them into prompt prototype. We then leverage contrastive learning to aggregate event mentions into prototypes and maintain these prototypes for few-shot ED. Furthermore, We explore diverse prompt formats for representing prompt prototypes and introduce a more comprehensive lexical prompt which improves the performance of few-shot ED. We conduct extensive experiments on the MAVEN corpus to reveal the effectiveness of the proposed framework compared to state-of-the-art methods.

Network security systems inspect packet payloads for signatures of attacks. These systems use regular expression matching at their core. Many techniques for implementing regular expression matching at line rate have been proposed. Solutions differ in the type of automaton used (i.e., deterministic versus non-deterministic) and in the configuration of implementation-specific parameters. While each solution has been shown to perform well on specific rule sets and traffic patterns, there has been no systematic comparison across a large set of solutions, rule sets and traffic patterns. Thus, it is extremely challenging for a practitioner to make an informed decision within the plethora of existing algorithmic and architectural proposals. Moreover, as multi-core processors are becoming popular, many parameters need to be tuned to maximize the multi-core potential. To address this problem, we present a comprehensive evaluation of a broad set of regular expression matching techniques. We consider both algorithmic and architectural aspects. Specifically, we explore the performance, area requirements, and power consumption of implementations targeting multi-core processors and FPGAs using rule sets of practical size and complexity. We present detailed performance results and specific guidelines for determining optimal configurations based on a simple evaluation of the rule set. These guidelines can help significantly when implementing regular expression matching systems in practice.

Deployment of innovative new networking services requires support by network providers. Since economic motivation plays an important role for network providers, it is critical that a network architecture intrinsically considers economic relationships. We present the design of a protocol that associates access to network services with economic contracts. We show how this protocol can be realized in fundamentally different ways, using out-of-band signaling and in-band signaling, based on two different prototype implementations. We present results that show the effectiveness of the proposed protocol and thus demonstrate a first step toward realizing an economy plane for the Internet.

Typical routing algorithms use a single criterion, such as hop count or link weight, to calculate paths. As the requirement of flexible routing arises, there are circumstances where multiple criteria are needed for routing. Though there are proposed solutions to the multi-criteria optimal path selection problem for quality-of-service routing, they usually combine all criteria into a single path optimization metric a priori. However, this approach is not feasible in scenarios where the path consumers' weightings of criteria is not known at compute time. Such circumstances require finding all the Pareto-optimal paths, i.e., all the paths that are not dominated by other paths. In this paper, we present the algorithmic foundations for efficiently computing Pareto-optimal paths. We present ParetoBFS, a variant of a breadth-first search that uses branch-and-bound techniques to find all the Pareto-optimal paths while effectively limiting the potentially very large search space. We present several sampling techniques to further increase the speed of the search while degrading the quality of the results only marginally. Our simulation results show that existing multi-criteria combinatorial optimization approaches can only search a small fraction of all the Pareto-optimal paths while ParetoBFS can obtain the whole path set in shorter time. We also present results from an implementation of ParetoBFS on a software-defined network prototype.

Network architectures for the future Internet envision a variety of novel network services for transmitting, processing, and storaging of data. These network services may involve costly resources that need to be allocated by a service provider. Thus, an important problem is to limit access to authorized users (e.g., Those who have paid for a particular network service). In addition, these resources need to be protected from denial-of-service attacks or attempts to circumvent this access control. Most existing authentication approaches are based on cryptographic techniques. However, the high computational cost of cryptographic operations makes these techniques unsuitable for the data plane of the network, where potentially every packet needs to be checked at Gigabit per second link rates. In this paper, we describe a novel design for data plane capabilities, called OrthCredential, that solves this problem. The main idea is to use a set of orthogonal sequences as credentials that can be verified easily to protect the data plane against various attacks. These orthogonal sequences can be constructed by a Hadamard transform. Our evaluation of a prototype implementation shows that 64-bit credentials only require less than 300 processor cycles for verification, much less than existing access control schemes such as HMAC. And it provides reasonable security properties (e.g., Less than 10 -- 8 probability of successful attack).

Network security systems inspect packet payloads for signatures of attacks. These systems use regular expression matching at their core. Many techniques for implementing regular expression matching at line rate have been proposed. Solutions differ in the type of automaton used (i.e., deterministic vs. non-deterministic) and in the configuration of implementation-specific parameters. While each solution has been shown to perform well on specific rule sets and traffic patterns, there has been no systematic comparison across a large set of solutions, rule sets and traffic patterns. Thus, it is extremely challenging for a practitioner to make an informed decision within the plethora of existing algorithmic and architectural proposals. To address this problem, we present a comprehensive evaluation of a broad set of regular expression matching techniques. We consider both algorithmic and architectural aspects. Specifically, we explore the performance, area requirements, and power consumption of implementations targeting processors and field programmable gate arrays using rule sets of practical size and complexity. We present detailed performance results and specific guidelines for determining optimal configurations based on a simple evaluation of the rule set. These guidelines can help significantly when implementing regular expression matching systems in practice.

Modern routers use high-performance multi-core packet processing systems to implement protocol operations and to forward traffic. As the diversity of protocols and the number of processor cores increases, it becomes increasingly difficult to manage these systems and ensure their correct operation at runtime. In particular, it is challenging to identify situations in which a part of processor cores behave incorrectly, either due to failure or due to malicious attacks. To address this problem, we present a novel approach to verifying correct operation of a packet processor by analyzing packet latency and throughput. This approach can treat the network processor as a “black box” and does not need to observe internal functionality. We show that processing time statistics are affected by system misbehavior and present an analytic model to quantify these effects. Our results show that the presented technique is an effective approach to provide an extra level of protection to packet processor systems.

The system architecture of a network system is of fundamental importance to the performance of computer networks. However, evaluating different architectural and algorithmic choices of network systems is usually difficult. This poster presents NetSysSim, a network system simulator implemented in SystemC. NetSysSim provides a hierarchical framework of different levels of abstraction. With SystemC's modular and timing based simulation, researchers can simulate any part of the system with the desired level of details. Our design of NetSysSim allows for performance evaluation of data plane architectures, including different queuing, scheduling or transmission schemes.

TCP and IP fragmentation can be used to evade signature detection at Intrusion Detection/Prevention System (IDS / IPS). Such fragments may arrive out-of-sequence to escape from being detected by the string matching algorithm of IDS / IPS. The common defense is buffering and reassembling packets. However, buffering of out-of-sequence packets can become impractical on high speed links due to limited fast memory capacity, especially when the concurrent flows are in large quantity, or extremely disordered in circumstances such as attacks. So such buffering strategy is vulnerable to memory exhausting denial of service (DoS). In this paper, AC-Suffix-Tree, a buffer free scheme for string matching is proposed, which detects patterns across out-of-sequence packets without buffering and reassembly. This novel algorithm associates the classical Aho-Corasick (AC) algorithm with a pattern suffix tree to search patterns with only the state numbers of AC automaton and suffix tree stored. It demands significantly less memory than buffering the packets themselves. Therefore the IDS can resist memory exhausting DoS attack. AC-Suffix-Tree consumes 1-2 orders of magnitude less memory than buffering the entire packet, and it has the same temporal complexity as AC algorithm when there are no out-of-sequence packets.

In this paper, we propose an extensible Open-Architecture Services platform (OASis) for high-performance network processing. OASis embraces recent advances of open technologies, including open source software, open system standards and open network architectures. Three programming models are proposed for target-specific processing modules: a multi-granularity packet processing model for network processing; a thread-isolated parallel programming model for service processing; and a message-based management model for centralized system administration. As an application example of OASis, a Unified Threat Management (UTM) prototype is implemented. This prototype provides multiple network security services, including stateful firewall, intrusion detection, and virus scanning. Experimental results show that, the OASis-UTM prototype can achieve 40Gbps stateful firewall performance together with 4--8Gbps intrusion detection and anti-virus performance on a 12U 14-slot ATCA platform.