Implicit hate speech has become a significant challenge for online platforms, as it often avoids detection by large language models (LLMs) due to its indirectly expressed hateful intent. This study identifies the limitations of LLMs in detecting implicit hate speech, particularly when disguised as seemingly harmless expressions in a rhetorical device. To address this challenge, we employ a Jailbreaking strategy and Energy-based Constrained Decoding techniques, and design a small model for measuring the energy of metaphorical rhetoric. This approach can lead to LLMs generating metaphorical implicit hate speech. Our research reveals that advanced LLMs, like GPT-4o, frequently misinterpret metaphorical implicit hate speech, and fail to prevent its propagation effectively. Even specialized models, like ShieldGemma and LlamaGuard, demonstrate inadequacies in blocking such content, often misclassifying it as harmless speech. This work points out the vulnerability of current LLMs to implicit hate speech, and emphasizes the improvements to address hate speech threats better.

Bragging, as a pervasive social-linguistic phenomenon, reflects complex human interaction patterns. However, the understanding and generation of appropriate bragging behavior in large language models (LLMs) remains underexplored. In this paper, we propose a comprehensive study that combines analytical and controllable approaches to examine bragging in LLMs. We design three tasks, bragging recognition, bragging explanation, and bragging generation, along with novel evaluation metrics to assess the models’ ability to identify bragging intent, social appropriateness, and account for context sensitivity. Our analysis reveals the challenges of bragging in the social context, such as recognizing bragging and responding appropriately with bragging in conversation. This work provides new insights into how LLMs process bragging and highlights the need for more research on generating contextually appropriate behavior in LLMs.

Few-Shot Document-Level Relation Extraction (FSDLRE) aims to develop models capable of generalizing to new categories with minimal support examples. Although Large Language Models (LLMs) demonstrate exceptional In-Context Learning (ICL) capabilities on many few-shot tasks, their performance on FSDLRE tasks remains suboptimal due to the significant gap between the task format and the intrinsic capabilities of language models, coupled with the complexity of ICL prompts for document-level text. To address these challenges, we introduce a novel meta-training approach for LLMs termed Prototype Tuning. We construct simulated episodes using data with relation types that do not overlap with the test corpus, fundamentally enhancing the ICL capabilities of LLMs in FSDLRE through meta-learning. To further enhance the effects of meta-learning, we innovatively integrate the concept of prototype into the fine-tuning process of LLMs. This involves aggregating entity pairs from support documents into prototypes within the prompts and altering the way of determining relation categories to identifying the closest prototype. Experimental results demonstrate that our LLMs trained with this approach outperform all baselines. Our proposed approach markedly improves the ICL capabilities of LLMs in FSDLRE and mitigates the impact of relation semantic discrepancies between the training corpus and the test corpus on model performance.

The proliferation of hate speech has caused significant harm to society. The intensity and directionality of hate are closely tied to the target and argument it is associated with. However, research on hate speech detection in Chinese has lagged behind, and existing datasets lack span-level fine-grained annotations. Furthermore, the lack of research on Chinese hateful slang poses a significant challenge. In this paper, we provide two valuable fine-grained Chinese hate speech detection research resources. First, we construct a Span-level Target-Aware Toxicity Extraction dataset (STATE ToxiCN), which is the first span-level Chinese hate speech dataset. Secondly, we evaluate the span-level hate speech detection performance of existing models using STATE ToxiCN. Finally, we conduct the first study on Chinese hateful slang and evaluate the ability of LLMs to understand hate semantics. Our work contributes valuable resources and insights to advance span-level hate speech detection in Chinese.

In this work, we tackle the challenge of multi-label emotion classification, where a sentence can simultaneously express multiple emotions. This task is particularly difficult due to the overlapping nature of emotions and the limited context available in short texts. To address these challenges, we propose an ensemble approach that integrates Pre-trained Language Models (BERT-based models) and Large Language Models, each capturing distinct emotional cues within the text. The predictions from these models are aggregated through a voting mechanism, enhancing classification accuracy. Additionally, we incorporate threshold optimization and class weighting techniques to mitigate class imbalance. Our method demonstrates substantial improvements over baseline models. Our approach ranked 4th out of 90 on the English leaderboard and exhibited strong performance in English in SemEval-2025 Task 11 Track A.

We propose a multilingual text processing framework that combines multilingual translation with data augmentation, QLoRA-based multi-model fine-tuning, and GLM-4-Plus-based ensemble classification. By using GLM-4-Plus to translate multilingual texts into English, we enhance data diversity and quantity. Data augmentation effectively improves the model’s performance on imbalanced datasets. QLoRA fine-tuning optimizes the model and reduces classification loss. GLM-4-Plus, as a meta-classifier, further enhances system performance. Our system achieved first place in three languages (English, Portuguese and Russian).

“目前基于预训练语言模型的司法领域低资源法律文书命名实体识别研究主要面临两个问题:(1)在低资源语言中,如维吾尔语,法律文书相关的语料极其有限,这种语料资源稀缺限制了基于预训练语言模型的训练和性能。(2)法律文书中使用的专业术语不仅复杂且特定,新的法律术语和概念的出现使得现有的模型难以适应。针对上述问题,本文基于多语言预训练模型mBERT,通过领域词汇扩充及模型微调的方法,提升了模型在维吾尔语法律文书命名实体识别任务的性能。本文首先整理并构建维吾尔语司法领域专业词汇列表,并将其添加到mBERT模型的词汇表中。随后,在人工标注的维吾尔语法律文书命名实体数据集UgLaw-NERD上进行模型微调,验证了该方法的有效性。实验结果表明,相比于仅使用mBERT进行微调的基线模型,融合领域词汇扩充的模型在命名实体识别任务上F1得分提升至89.72%,较基线提高了7.39%。此外,本文还探讨了不同领域词汇扩充量对模型命名实体识别性能的影响,结果显示,领域词汇扩充增强了预训练模型在处理维吾尔语任务中的表现。这些结论为其他低资源语言在司法领域开展基于预训练模型的自然语言处理研究提供了有益的参考。”

In recent years, with the vast and rapidly increasing amounts of spoken and textual data, Named Entity Recognition (NER) tasks have evolved into three distinct categories, i.e., text-based NER (TNER), Speech NER (SNER) and Multimodal NER (MNER). However, existing approaches typically require designing separate models for each task, overlooking the potential connections between tasks and limiting the versatility of NER methods. To mitigate these limitations, we introduce a new task named Integrated Multimodal NER (IMNER) to break the boundaries between different modal NER tasks, enabling a unified implementation of them. To achieve this, we first design a unified data format for inputs from different modalities. Then, leveraging the pre-trained MMSpeech model as the backbone, we propose an **I**ntegrated **M**ultimod**a**l **Ge**neration Framework (**IMAGE**), formulating the Chinese IMNER task as an entity-aware text generation task. Experimental results demonstrate the feasibility of our proposed IMAGE framework in the IMNER task. Our work in integrated multimodal learning in advancing the performance of NER may set up a new direction for future research in the field. Our source code is available at https://github.com/NingJinzhong/IMAGE4IMNER.

The Relational Triple Extraction (RTE) task is a fundamental and essential information extraction task. Recently, the table-filling RTE methods have received lots of attention. Despite their success, they suffer from some inherent problems such as underutilizing regional information of triple. In this work, we treat the RTE task based on table-filling method as an Object Detection task and propose a one-stage Object Detection framework for Relational Triple Extraction (OD-RTE). In this framework, the vertices-based bounding box detection, coupled with auxiliary global relational triple region detection, ensuring that regional information of triple could be fully utilized. Besides, our proposed decoding scheme could extract all types of triples. In addition, the negative sampling strategy of relations in the training stage improves the training efficiency while alleviating the imbalance of positive and negative relations. The experimental results show that 1) OD-RTE achieves the state-of-the-art performance on two widely used datasets (i.e., NYT and WebNLG). 2) Compared with the best performing table-filling method, OD-RTE achieves faster training and inference speed with lower GPU memory usage. To facilitate future research in this area, the codes are publicly available at https://github.com/NingJinzhong/ODRTE.

Chinese Named Entity Recognition (NER) has continued to attract research attention. However, most existing studies only explore the internal features of the Chinese language but neglect other lingual modal features. Actually, as another modal knowledge of the Chinese language, English contains rich prompts about entities that can potentially be applied to improve the performance of Chinese NER. Therefore, in this study, we explore the bilingual enhancement for Chinese NER and propose a unified bilingual interaction module called the Adapted Cross-Transformers with Global Sparse Attention (ACT-S) to capture the interaction of bilingual information. We utilize a model built upon several different ACT-Ss to integrate the rich English information into the Chinese representation. Moreover, our model can learn the interaction of information between bilinguals (inter-features) and the dependency information within Chinese (intra-features). Compared with existing Chinese NER methods, our proposed model can better handle entities with complex structures. The English text that enhances the model is automatically generated by machine translation, avoiding high labour costs. Experimental results on four well-known benchmark datasets demonstrate the effectiveness and robustness of our proposed model.

法律文本中包含的丰富信息可以通过结构化的实体关系三元组进行表示,便于法律知识的存储和查询。传统的流水线方法在自动抽取三元组时执行了大量冗余计算,造成了误差传播。而现有的联合学习方法无法适用于有大量重叠关系的法律文本,也并未关注语法结构信息对文本表示的增强,因此本文提出一种面向法律文本的实体关系联合抽取模型。该模型首先通过ON-LSTM注入语法信息,然后引入多头注意力机制分解重叠关系。相较于流水线和其他联合学习方法本文模型抽取效果最佳,在涉毒类法律文本数据集上抽取结果的F1值达到78.7%。

案件要素识别指将案件描述中重要事实描述自动抽取出来,并根据领域专家设计的要素体系进行分类,是智慧司法领域的重要研究内容。基于传统神经网络的文本编码难以提取深层次特征,基于阈值的多标签分类难以捕获标签间依赖关系,因此本文提出了基于预训练语言模型的多标签文本分类模型。该模型采用以Layer-attentive策略进行特征融合的语言模型作为编码器,使用基于LSTM的序列生成模型作为解码器。在“CAIL2019”数据集上进行实验,该方法比基于循环神经网络的算法在F1值上最高可提升7.6%,在相同超参数设置下比基础语言模型(BERT)提升约3.2%。

In recent years, the plentiful information contained in Chinese legal documents has attracted a great deal of attention because of the large-scale release of the judgment documents on China Judgments Online. It is in great need of enabling machines to understand the semantic information stored in the documents which are transcribed in the form of natural language. The technique of information extraction provides a way of mining the valuable information implied in the unstructured judgment documents. We propose a Legal Triplet Extraction System for drug-related criminal judgment documents. The system extracts the entities and the semantic relations jointly and benefits from the proposed legal lexicon feature and multi-task learning framework. Furthermore, we manually annotate a dataset for Named Entity Recognition and Relation Extraction in Chinese legal domain, which contributes to training supervised triplet extraction models and evaluating the model performance. Our experimental results show that the legal feature introduction and multi-task learning framework are feasible and effective for the Legal Triplet Extraction System. The F1 score of triplet extraction finally reaches 0.836 on the legal dataset.