From the viewpoint of high-generation citation networks, h indices can be extended to global metrics at both basic and aggregated levels. To accomplish this, three types of networks are designed: the paper-paper netwo...
详细信息
It is imperative to assure the security of smart contracts via intelligent vulnerability detection tools before deploying smart contracts on blockchains. The existing deep learning-based approaches fail to effectively...
详细信息
Event detection, an important research topic of information extraction, aims to automatically identify and classify event instances from the text. Previous studies have introduced methods combining syntactic informati...
详细信息
Event detection, an important research topic of information extraction, aims to automatically identify and classify event instances from the text. Previous studies have introduced methods combining syntactic information and graph convolutional networks into the field of event detection and verified their effectiveness. However, such methods often ignore the high-order information on the syntactic tree with noisy words, which limits their classification quality. In this paper, we propose a deep symmetric graph convolutional network to organically integrate high-order and low-order syntactic information to strengthen the semantic features of sentences. Specifically, we design a skip connection with attention gating mechanism, which selects valuable low-order syntactic information under the supervision of high-order syntactic information to strengthen the aggregation of high-order and low-order syntactic information. Then, a graph perturbation mechanism is proposed to discard noisy nodes on the syntactic graph to reduce the noisy information in the high-order syntactic information. We conducted extensive experiments on the widely used ACE 2005 benchmark, and the experimental results demonstrate that our method significantly outperforms state-of-the-art methods. Then, a graph perturbation mechanism is proposed to discard noisy nodes on the syntactic graph to reduce the noisy information in the high-order syntactic information. We conducted extensive experiments on the widely used ACE 2005 benchmark, and the experimental results demonstrate that our method significantly outperforms state-of-the-art methods. We conducted extensive experiments on the widely used ACE 2005 benchmark, and the experimental results demonstrate that our method significantly outperforms state-of-the-art methods. Then, a graph perturbation mechanism is proposed to discard noisy nodes on the syntactic graph to reduce the noisy information in the high-order syntactic information.
Recently, digital economy in social science has become increasingly significant. A well-constructed evaluation structure could generally and promptly describe and reveal its overall social digitalization development. ...
详细信息
Event Detection (ED) aims to recognize instances of specified types of event triggers in text. Existing graph neural network-based models have achieved promising progress to alleviate this problem by capturing differe...
详细信息
Event Detection (ED) aims to recognize instances of specified types of event triggers in text. Existing graph neural network-based models have achieved promising progress to alleviate this problem by capturing different orders of syntactic information, but they are limited by two issues. First, the long-range syntactic information between words is not fully exploited. Second, they ignore the semantic information provided by dependency labels which provide linguistic knowledge that is useful to ED. As a result, we proposed a label-enhanced dense graph convolutional network that employs dense connectivity and Graph Transformer Networks (GTN) to learn a flexible selection of edge types and composite relations between the words. Each layer can make use of the collective knowledge of dense blocks in order to model syntactic dependencies over long distances through dense connectivity. The proposed model achieves state-of-the-art performance for ED on common datasets after extensive experiments are conducted to show its advantages.
Question answering methods are well-known for leveraging data bias, such as the language prior in visual question answering and the position bias in machine reading comprehension (extractive question answering). Curre...
详细信息
Applying large language models (LLMs) to academic API usage shows promise in reducing researchers' efforts to seek academic information. However, current LLM methods for using APIs struggle with the complex API co...
详细信息
ISBN:
(纸本)9798400712456
Applying large language models (LLMs) to academic API usage shows promise in reducing researchers' efforts to seek academic information. However, current LLM methods for using APIs struggle with the complex API coupling commonly encountered in academic queries. To address this, we introduce SoAy, a solution-based LLM methodology for academic information seeking. SoAy enables LLMs to generate code for invoking APIs, guided by a pre-constructed API calling sequence referred to as a solution. This solution simplifies the model's understanding of complex API relationships, while the generated code enhances reasoning efficiency. LLMs are aligned with this solution-oriented, code-based reasoning method by automatically enumerating valid API coupling sequences and transforming them into queries and executable *** evaluate SoAy, we introduce SoAyBench, an evaluation benchmark accompanied by SoAyEval, built upon a cloned environment of APIs from AMiner. Experimental results demonstrate a 34.58-75.99% performance improvement compared to state-of-the-art LLM API-based baselines. All datasets, codes, tuned models, and deployed online services are publicly accessible at https://***/RUCKBReasoning/SoAy.
Structure from motion has attracted a lot of research in recent years, with new state-of-the-art approaches coming almost every year. One of its advantages over 3D reconstruction is that it can be used for any cameras...
详细信息
Computational Pathology (CPATH) offers the possibility for highly accurate and low-cost automated pathological diagnosis. However, the high time cost of model inference is one of the main issues limiting the applicati...
详细信息
Unsupervised person search aims to localize a particular target person from a gallery set of scene images without annotations, which is extremely challenging due to the unexpected variations of the unlabeled domains. ...
详细信息
暂无评论