Generative LLMs typically improve Named Entity Recognition (NER) performance through instruction tuning. They excel at generating entities by semantic pattern matching but lack an explicit, verifiable reasoning mechanism. This "cognitive shortcutting" leads to suboptimal performance and weak generalization, especially in zero-shot and low-resource scenarios where reasoning from limited contextual cues is crucial. To address this issue, a reasoning framework is proposed for NER, which shifts the extraction paradigm from implicit pattern matching to explicit reasoning. This framework consists of three stages: Chain of Thought (CoT) generation, CoT tuning, and reasoning enhancement. First, a dataset annotated with NER-oriented CoTs is generated, which contain task-relevant reasoning chains. Then, they are used to tune the NER model to generate coherent rationales before deriving the final answer. Finally, a reasoning enhancement stage is implemented to optimize the reasoning process using a comprehensive reward signal. This stage ensures explicit and verifiable extractions. Experiments show that ReasoningNER demonstrates impressive cognitive ability in the NER task, achieving competitive performance. In zero-shot settings, it achieves SoTA performance, outperforming GPT-4 by 12.3 percentage points on the F1 score. Analytical results demonstrate its great potential to advance research in reasoning-oriented information extraction.

The proliferation of textual information in the digital age has made multi-document summarization (MDS) a critical tool for efficient information access. Traditional MDS approaches often struggle with input length constraints, redundancy, and coherence. In this work, we propose a novel sequential summarization paradigm: instead of generating a summary for the entire document set simultaneously, we iteratively summarize documents by integrating the current document with the previously generated summary. This progressive strategy enables incremental synthesis and alleviates token-length bottlenecks in large language models. To mitigate error accumulation and preserve factual consistency, we introduce a modular framework based on a fine-tuned pre-trained language model, augmented with lightweight auxiliary models for content selection and verification. Experiments on eight public datasets spanning news, scientific, legal, and clinical domains show up to +3 ROUGE-L and +5 BLEU over strong baselines and consistently improves BERTScore and FactCC by 1–3 points. Ablation studies validate the contribution of each module. Our findings highlight the effectiveness and scalability of iterative multi-document generation in producing coherent, concise, and factually grounded summaries.

Fuzzy logic-based graph neural networks (FL-GNN) have recently garnered growing attention in node classification, which aims to enhance the ability of GNN in modeling uncertain relationships between nodes. However, existing FL-GNN typically assume that nodes in the source domain (training set) and target domain (test set) follow the identical data distribution and class sets. Real-world scenarios often exhibit significant distribution shifts and target domain even contains classes that were not present in the source domain, termed open-set cross-domain node classification (OSCD-NC), which seriously damages their superior performance. Thus, how to leverage the strong uncertain knowledge representation capacity of FL-GNN to learn a well-defined boundary between seen and unseen classes for improving OSCD-NC performance remains an open and underexplored research problem. In this paper, we propose an effective domain-adaptive fuzzy graph diffusion network (DFGDN) for OSCD-NC. Specifically, with the help of a fuzzy adjacency matrix, fuzzy graph diffusion networks are proposed to generate robust fuzzy node representations by adaptively enhancing feature collaboration between low-pass and high-pass graph filters. Then, a peer ($M$+1)-class classifier is introduced to learn a rough class boundary by measuring their class prediction probability difference for target domain. After that, the ($M$+1)-means clustering and decoder modules are simultaneously designed to discover more supervision guidance from target domain for learned class boundary optimization. Finally, we jointly optimize the above modules in an adversarial manner via classification loss, classifier discrepancy loss and mean squared error loss, which further improves the accuracy of the learned class boundary by pulling seen nodes from the source domain and target domain closer, and pushing unseen nodes away. Extensive experiments on three cross-domain data pairs and various openness rates demonstrate the effectiveness of the proposed DFGDN framework.

Multiple-choice questions (MCQs) are a widely used and vital assessment format for evaluating large language models (LLMs). This study reveals that LLMs are susceptible to “cognitive load” caused by distractor options in MCQs, leading to excessive attention to distractors and consequent vacillation between correct and incorrect options. To mitigate this cognitive burden, we introduce a novel reasoning prompt strategy, called EoT, which effectively reduces cognitive load by steering the model’s attention away from erroneous options. This enables the model to focus more effectively on reasonable answers. Additionally, by documenting the elimination process, EoT enhances the transparency and interpretability of the model’s reasoning. Experimental results demonstrate that EoT, as a plug-and-play approach, significantly reduces cognitive load and improves performance, showcasing its potential to enhance both the accuracy and interpretability of LLMs.

Multi-label text classification (MLTC) refers to that each document is associated with more than one label at the same time, which attract much attention from researchers in both academia and industry. Existing methods have difficulties in determining label-related components from documents, which cannot effectively establish the association between textual features and label information. In fact, there are some label information, such as label semantic information and co-occurrence relations among labels, could be used to improve the performance on multi-label text classification. In this paper, we propose a label information aware model to utilize these information. Our model makes use of label semantic information to determine label-related components from textual features for obtaining the label-specific textual presentation for each sample, and then take advantages of co-occurrence relations among labels to construct interaction among label-specific textual presentation. The superiority of our model has been proved through comparing our method with several existing models on two datasets.

Abstract Prompt-tuning has been successfully applied to support classification tasks in natural language processing and has achieved promising performance. The main characteristic of prompt-tuning based classification is to verbalize class labels and predict masked tokens like a cloze-like task. It has the advantage to make use of knowledge in pre-trained language models (PLMs). Because prompt templates are manually designed, they are more prone to overfitting. Furthermore, traditional prompt templates are appended in the tail of an original sentence. They are far from some semantic units in a sentence. It is weak to decode semantic information of an input relevant to PLMs. To aggregate more semantic information from PLMs for masked token prediction, we propose an annotation-aware prompt-tuning model for relation extraction. In our method, entity type representations are used as entity annotations. They are implanted near the site of entities in a sentence for decoding semantic information of PLMs. It is effective to make full use of knowledge in PLMs for relation extraction. In the experiment section, our method is validated on the Chinese literature text and SemEval 2010 Task datasets and achieves 89.3% and 90.6% in terms of F1-score, respectively. It achieves the state-of-the-art performance on two public datasets. The result further demonstrates the effectiveness of our model to decode semantic information in PLMs.

In this paper, a deep multi-kernel clustering network, named DMKCN, is proposed to learn a high-quality and structurally separable kernel representation for the clustering task. Specifically, a multi-kernel learner is proposed to choose a suitable kernel function by learning a suitable combination of kernel functions automatically. A kernel-aid encoder module, consisting of a series of multi-kernel learners, is proposed to learn the structurally separable kernel representation. Besides, a dual self-supervised mechanism, consisting of a kernel self-supervised strategy and a representation self-supervised strategy, is designed to uniformly optimize the kernel representation learning and structural partition. The kernel self-supervised strategy is developed to supervise the multi-kernel learners with the consideration of an objective of clustering task, the representation self-supervised strategy is developed to guide the optimization of kernel representation learning by reconstructing the raw data. Extensive experiments on six real-world datasets demonstrate the outstanding performance of our proposed DMKCN.

Prompt-tuning models output relation types as verbalised-type tokens instead of predicting the confidence scores for each relation type. However, existing prompt-tuning models cannot perceive named entities of a relation instance because they are normally implemented on raw input that is too weak to encode the contextual features and semantic dependencies of a relation instance. This study proposes a cue prompt adapting (CPA) model for relation extraction (RE) that encodes contextual features and semantic dependencies by implanting task-relevant cues in a sentence. Additionally, a new transformer architecture is proposed to adapt pre-trained language models (PLMs) to perceive named entities in a relation instance. Finally, in the decoding process, a goal-oriented prompt template is designed to take advantage of the potential semantic features of a PLM. The proposed model is evaluated using three public corpora: ACE, ReTACRED, and Semeval. The performance achieves an impressive improvement, outperforming existing state-of-the-art models. Experiments indicate that the proposed model is effective for learning task-specific contextual features and semantic dependencies in a relation instance.

Span classification is a popular method for nested named entity recognition. To recognise full-span named entities, span-based models should enumerate and verify all possible entity spans in a sentence, which leads to serious problems regarding computational complexity and data imbalance. In this study, we propose a boundary regression model to support full-span named entity recognition, where a regression operation is adopted to refine spatial locations of entity spans in a sentence. Therefore, instead of exhaustively enumerating all possible spans, we need only verify a small number of them. Span boundaries are regressed to find all possible named entities in a sentence. Furthermore, for a better representation of long-named entities, a multi-granule sentence representation is adopted to encode semantic features with different semantic granularities. In our experiments, even enumerating a small number of entity spans, our model still has competitive performance, achieving 87.35% and 80.85% F1 scores on the ACE2005 and GENIA datasets. Analytical experiments show that our model is able to find all named entities in a sentence without exhaustively verifying all possible entity spans. It is effective in mitigating the computational complexity and data imbalance problems in full-span named entity recognition.

Nested named entities (nested NEs) refer to the situation where one named entity is included or nested within another named entity, which cannot be recognized by the traditional sequence labeling methods. Recently, span-based methods have become the mainstream methods for nested Named Entity Recognition (nested NER). The fundamental concept behind this method is to enumerate nearly all potential spans as entity mentions and subsequently classify them. However, span-based methods independently classify spans without considering the semantic relations among them, which negatively impacts the span representation. To address the issue, we propose a novel deep learning architecture for nested NER that explores interactive and contrastive relations among spans. Specifically, we design a scale transformation mechanism that embeds geometric information into span representations, which enhances the model's ability to encode interactive relations between spans. Additionally, we introduce a supervised contrastive learning loss that pulls apart highly overlapping spans in the embedding space to encode the contrastive relations. Experiments show that our method achieves state-of-the-art or competitive performance on three publicly nested NER datasets, thus validating its effectiveness.

By network security threat intelligence analysis based on a security knowledge graph (SKG), multi-source threat intelligence data can be analyzed in a fine-grained manner. This has received extensive attention. It is difficult for traditional named entity recognition methods to identify mixed security entities in Chinese and English in the field of network security, and there are difficulties in accurately identifying network security entities because of insufficient features extracted. In this paper, we propose a novel FT-CNN-BiLSTM-CRF security entity recognition method based on a neural network CNN-BiLSTM-CRF model combined with a feature template (FT). The feature template is used to extract local context features, and a neural network model is used to automatically extract character features and text global features. Experimental results showed that our method can achieve an F-score of 86% on a large-scale network security dataset and outperforms other methods.

Abstract It is a challenging task to discover information from a large amount of data in an open domain.1In this paper, an event network framework is proposed to address this challenge. It is in fact an empirical construct for exploring open information, composed of three steps: document event detection, event network construction and event network analysis. First, documents are clustered into document events for reducing the impact of noisy and heterogeneous resources. Secondly, linguistic units (e.g., named entities or entity relations) are extracted from each document event and combined into an event network, which enables content-oriented retrieval. Then, in the final step, techniques such as social network or complex network can be applied to analyze the event network for exploring open information. In the implementation section, we provide examples of exploring open information via event network.

Chinese is an ancient hieroglyphic. It is inat-tentive to structure. Therefore, segmenting and parsing Chinese are more difficult and less accurate. In this paper, we propose an Omni-word feature and a soft constraint method for Chinese relation extraction. The Omni-word feature uses every potential word in a sentence as lexicon feature, reducing errors caused by word segmentation. In order to utilize the structure information of a relation instance, we discuss how soft constraint can be used to cap-ture the local dependency. Both Omni-word feature and soft constraint make a better use of sentence information and minimize the in-fluences caused by Chinese word segmenta-tion and parsing. We test these methods on the ACE 2005 RDC Chinese corpus. The re-sults show a significant improvement in Chi-nese relation extraction, outperforming other methods in F-score by 10 % in 6 relation types and 15 % in 18 relation subtypes. 1