Multi-Modal Military Event Extraction Based on Knowledge Fusion

Event extraction stands as a significant endeavor within the realm of information extraction,aspiring to automatically extract structured event information from vast volumes of unstructured text.Extracting event elements from multi-modal data remains a challenging task due to the presence of a large number of images and overlapping event elements in the data.Although researchers have proposed various methods to accomplish this task,most existing event extraction models cannot address these challenges because they are only applicable to text scenarios.To solve the above issues,this paper proposes a multi-modal event extraction method based on knowledge fusion.Specifically,for event-type recognition,we use a meticulous pipeline approach that integrates multiple pre-trained models.This approach enables a more comprehensive capture of the multidimensional event semantic features present in military texts,thereby enhancing the interconnectedness of information between trigger words and events.For event element extraction,we propose a method for constructing a priori templates that combine event types with corresponding trigger words.This approach facilitates the acquisition of fine-grained input samples containing event trigger words,thus enabling the model to understand the semantic relationships between elements in greater depth.Furthermore,a fusion method for spatial mapping of textual event elements and image elements is proposed to reduce the category number overload and effectively achieve multi-modal knowledge fusion.The experimental results based on the CCKS 2022 dataset show that our method has achieved competitive results,with a comprehensive evaluation value F1-score of 53.4%for the model.These results validate the effectiveness of our method in extracting event elements from multi-modal data.

Biomedical Event Extraction Using a New Error Detection Learning Approach Based on Neural Network

Supervised machine learning approaches are effective in text mining,but their success relies heavily on manually annotated corpora.However,there are limited numbers of annotated biomedical event corpora,and the available datasets contain insufficient examples for training classifiers;the common cure is to seek large amounts of training samples from unlabeled data,but such data sets often contain many mislabeled samples,which will degrade the performance of classifiers.Therefore,this study proposes a novel error data detection approach suitable for reducing noise in unlabeled biomedical event data.First,we construct the mislabeled dataset through error data analysis with the development dataset.The sample pairs’vector representations are then obtained by the means of sequence patterns and the joint model of convolutional neural network and long short-term memory recurrent neural network.Following this,the sample identification strategy is proposed,using error detection based on pair representation for unlabeled data.With the latter,the selected samples are added to enrich the training dataset and improve the classification performance.In the BioNLP Shared Task GENIA,the experiments results indicate that the proposed approach is competent in extract the biomedical event from biomedical literature.Our approach can effectively filter some noisy examples and build a satisfactory prediction model.

Joint Event Extraction Based on Global Event-Type Guidance and Attention Enhancement

Event extraction is one of the most challenging tasks in information extraction.It is a common phenomenon where multiple events exist in the same sentence.However,extracting multiple events is more difficult than extracting a single event.Existing event extraction methods based on sequence models ignore the interrelated information between events because the sequence is too long.In addition,the current argument extraction relies on the results of syntactic dependency analysis,which is complicated and prone to error trans-mission.In order to solve the above problems,a joint event extraction method based on global event-type guidance and attention enhancement was proposed in this work.Specifically,for multiple event detection,we propose a global-type guidance method that can detect event types in the candidate sequence in advance to enhance the correlation information between events.For argument extraction,we converted it into a table-flling problem,and proposed a table-flling method of the attention mechanism,that is simple and can enhance the correlation between trigger words and arguments.The experimental results based on the ACE 2005 dataset showed that the proposed method achieved 1.6%improvement in the task of event detection,and obtained state-of-the-art results in the argument extraction task,which proved the effectiveness of the method.

Automatic Event Trigger Word Extraction in Chinese Event

As a basic unit of knowledge representation and an important means for information organization, event has drawn growing number of people’s attention, the research of event identification and extraction in natural language processing field is an important research topic in information extraction area, the recognition and extraction of event trigger word plays a decisive role in event identification and extraction. In this paper, the authors make experiment in Chinese Event Corpus CEC, and present a method of extracting event trigger word automatically that combines extended trigger word table and machine learning. The experiment result shows that the F-score of extracting event trigger word. can reach 71.2% by using this method.

KeyEE:Enhancing Low-Resource Generative Event Extraction with Auxiliary Keyword Sub-Prompt

Event Extraction(EE)is a key task in information extraction,which requires high-quality annotated data that are often costly to obtain.Traditional classification-based methods suffer from low-resource scenarios due to the lack of label semantics and fine-grained annotations.While recent approaches have endeavored to address EE through a more data-efficient generative process,they often overlook event keywords,which are vital for EE.To tackle these challenges,we introduce KeyEE,a multi-prompt learning strategy that improves low-resource event extraction by Event Keywords Extraction(EKE).We suggest employing an auxiliary EKE sub-prompt and concurrently training both EE and EKE with a shared pre-trained language model.With the auxiliary sub-prompt,KeyEE learns event keywords knowledge implicitly,thereby reducing the dependence on annotated data.Furthermore,we investigate and analyze various EKE sub-prompt strategies to encourage further research in this area.Our experiments on benchmark datasets ACE2005 and ERE show that KeyEE achieves significant improvement in low-resource settings and sets new state-of-the-art results.

The Adverse Event Profile in Patients Treated with Transferon^TM(Dialyzable Leukocyte Extracts): A Preliminary Report

Background: Dialyzable leukocyte extracts (DLE) are heterogeneous mixtures of peptides less than 10 kDa in size that are used as immunomodulatory adjuvants in immune-mediated diseases. TransferonTM is DLE manufactured by National Polytechnic Institute (IPN), and is registered by Mexican health-regulatory authorities as an immunomodulatory drug and commercialized nationally. The proposed mechanism of action of TransferonTM is induction of a Th1 immunoregulatory response. Despite that it is widely used, to date there are no reports of adverse events related to the clinical safety of human DLE or TransferonTM. Objective: To assess the safety of TransferonTM in a large group of patients exposed to DLE as adjuvant treatment. Methods: We included in this study 3844 patients from our Clinical Immunology Service at the Unit of External Services and Clinical Research (USEIC), IPN. Analysis was performed from January 2014 to November 2014, searching for clinical adverse events in patients with immune-mediated diseases and treated with TransferonTM as an adjuvant. Results: In this work we observed clinical nonserious adverse events (AE) in 1.9% of patients treated with TransferonTM (MD 1.9, IQR 1.7 - 2.0). AE were 2.8 times more frequently observed in female than in male patients. The most common AE were headache in 15.7%, followed by rash in 11.4%, increased disease-related symptomatology in 10%, rhinorrhea in 7.1%, cough in 5.7%, and fatigue in 5.7% of patients with AE. 63% of adverse event presentation occurred from day 1 to day 4 of treatment with TransferonTM, and mean time resolution of adverse events was 14 days. In 23 cases, the therapy was stopped because of adverse events and no serious adverse events were observed in this study. Conclusion: TransferonTM induced low frequency of nonserious adverse events during adjuvant treatment. Further monitoring is advisable for different age and disease groups of patients.

Combing Type-Aware Attention and Graph Convolutional Networks for Event Detection

Event detection(ED)is aimed at detecting event occurrences and categorizing them.This task has been previously solved via recognition and classification of event triggers(ETs),which are defined as the phrase or word most clearly expressing event occurrence.Thus,current approaches require both annotated triggers as well as event types in training data.Nevertheless,triggers are non-essential in ED,and it is time-wasting for annotators to identify the“most clearly”word from a sentence,particularly in longer sentences.To decrease manual effort,we evaluate event detectionwithout triggers.We propose a novel framework that combines Type-aware Attention and Graph Convolutional Networks(TA-GCN)for event detection.Specifically,the task is identified as a multi-label classification problem.We first encode the input sentence using a novel type-aware neural network with attention mechanisms.Then,a Graph Convolutional Networks(GCN)-based multilabel classification model is exploited for event detection.Experimental results demonstrate the effectiveness.

基于任务转化的事件抽取通用框架

论元分散、多重事件和论元重叠是事件抽取长期面临的挑战,许多任务还存在触发词或位置标注缺失的情况。对此,提出一种基于任务转化的事件抽取通用框架,包括任务转化、关系抽取和事件预测3个模块。有触发词和无触发词的标注事件分别被转化为不同形式的关系三元组;这些三元组将与原始文本一起作为关系抽取模型的训练数据;事件预测时先从输入文本中抽取三元组,再将它们还原为目标事件。对于训练语料中位置标注缺失的情况,设计了基于最短距离的论元定位算法。该框架在ChFinAnn数据集上取得81.6%的平均F1值,在DuEE-Fin数据集上的F1值为72.04%(在线提交结果),均达到目前的SOTA水平。实验结果表明,该框架不仅可以显著提高事件抽取效果,而且具有广泛的适应能力。

基于多粒度阅读器和图注意力网络的文档级事件抽取

文档级事件抽取面临论元分散和多事件两大挑战,已有工作大多采用逐句抽取候选论元的方式,难以建模跨句的上下文信息。为此,提出了一种基于多粒度阅读器和图注意网络的文档级事件抽取模型,采用多粒度阅读器实现多层次语义编码,通过图注意力网络捕获实体对之间的局部和全局关系,构建基于实体对相似度的剪枝完全图作为伪触发器,全面捕捉文档中的事件和论元。在公共数据集ChFinAnn和DuEE-Fin上进行了实验,结果表明提出的方法改善了论元分散问题,提升了模型事件抽取性能。

地震事件分类识别软件

非天然地震事件分类是地震监测业务部门的日常工作之一。本研究主要针对地震、爆炸和矿震的分类问题,在地震波数据处理、特征提取和人工智能综合分类的研究基础上,基于Qt开发框架,结合Python、 Matlab等多种编程语言,开发了一个具有良好的可移植性和可扩展性、具有自主知识产权的地震分类识别软件。该软件可以部署在不同操作系统上,由七个模块组成:地震数据导入模块、数据处理模块、特征提取模块、综合分类模块、特征分析模块、当量估算模块和结果分析模块。软件集成了多种时频特征提取技术和人工智能分类方法,形成了较为完整的地震类型判定流程。软件内置的地震事件分类模型准确率高于90%,适用范围较广,已推广应用于多个地震监测部门,并取得了较好的应用成果,提高了对非天然地震的快速分析能力。

基于自注意力机制的中文金融事件元素抽取

针对中文金融事件元素抽取任务中多个代词指代同一个元素的问题,提出了基于自注意力机制的事件元素抽取模型。该模型在预处理阶段融入金融事件领域知识与事件类型知识,使得预训练模型可以根据事件类型信息获得更可靠的事件元素表示;然后,使用多头注意力机制挖掘新闻上下文不同元素指代词的指代含义,做到重叠元素间的指代消解;最后,使用双向长短期记忆网络与条件随机场挖掘新闻长文本的上下文特征表示,实现事件元素抽取。构建了中文金融事件语料库,通过与主流模型的对比实验验证了该模型的有效性。

文档级事件抽取反向推理模型

事件抽取旨在从非结构化文本中检测事件类型并抽取事件要素。现有方法在处理文档级文本时仍存在局限性。这是因为文档级文本可能由多个事件组成,并且构成某一事件的事件要素通常分散在不同句子中。为应对上述挑战,提出了一种文档级事件抽取反向推理模型(reverse inference model for document-level event extraction,RIDEE)。基于无触发词的设计,将文档级事件抽取转化为候选事件要素抽取和事件触发推理两个子任务,并行式抽取事件要素并检测事件类型。此外,设计了一种用于存储历史事件的事件依赖池,使得模型在处理多事件文本时可以充分利用事件之间的依赖关系。公开数据集上的实验结果表明,与现有事件抽取模型相比,RIDEE在进行文档级事件抽取时具有更优的性能。

基于Fin-BERT的中文金融领域事件抽取方法

事件抽取旨在从海量的非结构化的事件相关文本中抽取出人类感兴趣的内容,目前现有的事件抽取方法大多数基于通用语料,很少考虑到领域内的先验知识,并且现有的方法大多数不能很好地处理同一文档包含多个事件的情况,面对存在较多负面样例的测试也表现不佳。针对上述问题提出了一种基于Fin-BERT(financial bidirectional encoder representation from Transformers)和PTPCG(pseudo-trigger-aware pruned complete graph)的模型FinPTPCG,该方法充分利用Fin-BERT预训练模型的表达能力,在编码阶段融入领域内的先验知识,并且在事件检测模块采用多个二元分类器叠加的方式,保证模型可以有效识别一篇文档内存在多事件的情况并筛除掉负面样例,抽取实体之后将实体连接成完全图并通过计算相似度矩阵进行剪枝,通过选择伪触发器解决无标注触发词的问题,最后接入事件分类器实现事件抽取。该方法在ChFinAnn和Duee-fin数据集上事件抽取任务的F1值相比于基线方法分别取得了0.7个百分点和3.7个百分点的提升。

基于联合学习的语言粒度融合的重叠事件抽取方法

事件抽取是一项重要的信息抽取任务,现有的事件抽取方法大多假设一个句子中仅出现一个事件,然而,在真实的场景下,重叠事件是难以避免的。文中提出了一种基于联合学习的语言粒度融合的重叠事件抽取方法。该方法设计了基于token数目逐层递增和逐层递减的策略,对不同语言粒度的片段进行表示,在此基础上,构建了渐进式语言粒度融合的句子表示。通过引入事件信息感知,建立了基于门控机制的语言粒度和事件信息融合的句子表示。最后,通过联合学习词间的片段关系和角色关系,实现对事件触发词、论元、事件类型和论元角色的判别。在FewFC和DuEE1.0-1数据集上进行了实验,所提LGFEE模型在事件类型判别任务上的F1值分别提高了0.8%和0.6%,在触发词识别、论元识别、论元角色分类任务上也获得了较高的召回率和F1值,验证了其有效性。

篇章级文本的事件图谱构建与事件抽取方法

针对篇章级事件抽取任务主要存在的以下3个挑战:模型复杂、事件论元分散以及多事件抽取,提出一种基于异质交互图和有序树的篇章级事件抽取方法DTHIGN。支持以解耦实体抽取部分的方式精简模型的参数;构建面向触发词的篇章级异质交互图,更全面地对语义进行建模;支持使用基于统一事件模板的树形事件论元抽取方式获得事件论元结果。实验结果表明,该方法能够获得较好的效果并显著降低模型的训练难度。

基于高置信度伪标签数据选择算法的临床事件抽取方法

【目的】事件抽取是构建高质量事件知识图谱的前提。临床事件抽取过程中事件元素存在依赖关系,现有方法无法准确识别事件元素并组合为事件,且现有临床事件标记数据较少,给事件抽取任务带来了极大的挑战。【方法】将临床事件抽取建模为实体识别模型,提出一种融合多特征的中文医学事件抽取方法:BERT-MCRF.该方法使用BERT构建模型的嵌入和特征提取部分,在CRF层加入多个字的滑动窗口特征,然后将BERT-MCRF作为半监督实验的基实验,提出一种高置信度伪标签数据选择算法作为筛选数据的条件,得到较高质量的300条数据与原始数据合并,最终构建了1700条语料,并重新训练模型。【结果】BERT-MCRF模型在3种属性实体上的整体F1值达到80.21%,比经典的BiLSTM-CRF模型提升15.11%;通过半监督思路重新训练的模型最终F1值达到81.56%,较原始BERT-MCRF提升了1.35%.

基于知识提示微调的事件抽取方法

事件抽取是信息抽取中的一个重要研究热点,旨在通过识别和分类事件触发词和论元,从文本中抽取出事件结构化信息。传统的方法依赖于复杂的下游网络,需要足够的训练数据,在数据稀缺的情况下表现不佳。现有研究利用提示学习,在事件抽取上取得一定的研究成果,但依赖手工构建,且只依靠预训练模型已有的知识,缺乏事件特有的知识。因此本文提出一种基于知识提示微调的事件抽取方法。该方法采用条件生成的方式,在现有预训练语言模型知识的基础上,注入事件信息以提供论元关系约束,并采用提示微调策略对提示进行优化。大量实验结果表明,相较于传统基线方法,该方法在触发词抽取上优于基线方法,并在小样本下达到最好的效果。

低资源场景事件抽取研究综述

事件抽取作为信息抽取的任务之一,旨在从非结构化文本中抽取出结构化事件信息。当前基于机器学习和深度学习的自动化信息抽取方法过度依赖于标注数据,而大多数领域的标准数据集规模很小且分布不均匀,因此低资源场景成为了限制自动化信息抽取性能的瓶颈。虽然,近年来众多学者针对低资源场景进行了深入研究,并取得了许多显著的成果,但目前针对该场景下的事件抽取研究综述比较缺乏。文中对现有的学术成果进行了较为全面的总结分析,首先介绍了相关任务的定义,并将低资源场景事件抽取任务分为3类;其次围绕此分类重点阐述了6种相关技术方法,包括基于迁移学习、基于提示学习、基于无监督学习、基于弱监督学习、基于数据与辅助知识增强、基于元学习的方法,并指出了当前方法的不足和未来改进的方向;然后介绍了相关数据集及评价指标,并对典型技术方法的实验结果进行了总结分析;最后从全局角度总结分析了当前低资源场景事件抽取工作面临的挑战及未来研究的趋势。

基于ABBSAC模型的中文事件抽取方法

事件抽取作为信息抽取的重要一环,是非结构化文本转化为有价值的结构化文本的主要方式。针对目前事件抽取模型普遍训练时间长、模型体量大等问题,提出了一个基于ABBSAC的中文事件抽取模型。通过ALBERT预训练模型缩减模型体量,采用BiSRU++捕捉文本内部关联信息,并融合注意力机制提升模型精度,最后以CRF的输出作为抽取结果。基于新浪新闻自主构建了语料集,进行了对比实验。在获得较高准确率、召回率以及F1值的基础上,该模型训练速度提高了约10%,模型参数量裁剪了约82%,证明了所提模型的先进性。同时,在ACE05和DUEE基准测评数据集上,与前沿方法相比较,将触发词抽取的F1值分别提升了1.7%、0.3%,将论元角色抽取的F1值分别提升了5.4%、0.1%,有效提升了中文事件抽取任务的效能。

基于大语言模型隐含语义增强的细粒度虚假新闻检测方法

随着生成式人工智能技术的发展,许多领域都得到了帮助与发展,但与此同时虚假信息的构建与传播变得更加简单,虚假信息的检测也随之难度增加.先前的工作主要聚焦于语法问题、内容煽动性等方面的特点,利用深度学习模型对虚假新闻内容进行建模.这样的方式不仅缺乏对内容本身的判断,还无法回溯模型的判别原因.针对上述问题提出一种基于大语言模型隐含语义增强的细粒度虚假新闻检测方法.该方法充分挖掘并利用了现有的生成式大语言模型所具有的总结与推理能力,按照主干事件、细粒度次要事件和隐含信息推理的顺序进行层级式推导,逐步判别新闻的真实性.通过分解任务的方式,该方法最大程度发挥了模型的能力,提高了对虚假新闻的捕获能力,同时该方法也具有一定的可解释性,能够为检测提供判别依据.