基于实体活跃度及复制生成的时序知识图谱推理

现有时序知识图谱推理主要是基于静态知识图谱的推理方法,通过知识图谱的结构特征挖掘潜在的语义信息和关系特征,忽略了实体时序信息的重要性,因此提出一种基于实体活跃度及复制生成机制的时序知识图谱推理方法(EACG)。首先,通过改进的图卷积神经网络对多关系实体建模,有效挖掘知识图谱的潜在语义信息和结构特征。其次,时序编码器基于实体活跃度学习实体的时序特征。最后,使用复制生成机制进一步学习知识图谱的历史信息,提升对时序数据建模的能力。在时序知识图谱数据集ICEWS14、ICEWS05-15、GDELT上推理的实验结果表明,EACG在MRR评估指标中分别优于次优方法2%、10%和5%。     

时序知识图谱表示与推理的研究进展与趋势

知识图谱作为近年来人工智能领域的一大热点研究方向, 已应用于现实中多个领域. 但是随着知识图谱应用场景日益多样化, 人们逐渐发现不随着时间改变而更新的静态知识图谱不能完全适应知识高频更新的场景. 为此, 研究者们提出时序知识图谱的概念, 一种包含时间信息的知识图谱. 对现有所有时序知识图谱表示与推理模型进行整理, 并归纳和建立一个表示与推理模型理论框架. 然后基于此对当前时序表示推理研究进展进行简要介绍分析和未来趋势预测, 以期望帮助研究者开发设计出更为优异的模型.     

融合语义信息的知识图谱多跳推理模型

多跳推理模型在知识图谱中充分挖掘和利用实体间的多步关系,组成路径信息,完成知识推理,然而,目前的稀疏知识图谱多跳推理模型大多存在数据稀少及推理路径可靠性较低等问题.为了解决该问题,文中提出融合语义信息的知识图谱多跳推理模型.首先,将知识图谱中的实体和关系嵌入向量空间,作为强化学习训练的外部环境.然后,利用查询关系和推理路径的语义信息,选择相似度最高的(关系,实体)对扩充智能体进行路径搜索的动作空间,以此弥补推理过程中数据稀少的不足.最后,使用推理路径和查询关系的语义相似度评价推理路径的可靠性,并作为奖励函数反馈给智能体.在多个公开稀疏数据集上的实验表明,文中模型明显提升推理性能.     

时态知识图谱的推理研究综述

随着社交网络、物端感知等技术快速发展,网络空间中涌现了大量的交互、话题、事件、新闻等数据,蕴含大量动态演化、强时效性的知识.较于忽略知识中时间信息的传统知识图谱,时态知识图谱通过建模知识的时效性以描述动态变化的现实世界,为时间紧耦合的应用提供有效支持.然而,时态知识图谱无法确保涵盖全量知识,知识的缺失严重影响应用性能,需要推理模型自动挖掘新的知识,以解释事物的历史状态,预测未来发展趋势并描述演化规律.由于实际应用的迫切需要,近年来,时态知识图谱的推理研究工作层出不穷,逐渐引起学术界和工业界的广泛关注.本文对近年来时态知识图谱的推理工作进行全面介绍和总结.首先,介绍了时态知识图谱的推理相关概念与问题描述;其次,介绍了面向补全任务的推理模型与面向预测任务的推理模型,对其进行比较分析;之后总结了时态知识图谱推理的数据集、推理任务、相关指标以及应用场景;最后展望时态知识图谱推理的未来研究趋势.综上,本文致力于为时态知识图谱的推理领域研究人员提供具有价值的参考,以推动该领域进一步发展.     

知识图谱可解释推理研究综述

面向知识图谱的知识推理旨在通过已有的知识图谱事实,去推断新的事实,进而实现知识库的补全.近年来,尽管基于分布式表示学习的方法在推理任务上取得了巨大的成功,但是他们的黑盒属性使得模型无法为预测出的事实做出解释.所以,如何设计用户可理解、可信赖的推理模型成为了人们关注的问题.从可解释性的基本概念出发,系统梳理了面向知识图谱的可解释知识推理的相关工作,具体介绍了事前可解释推理模型和事后可解释推理模型的研究进展;根据可解释范围的大小,将事前可解释推理模型进一步细分为全局可解释的推理和局部可解释的推理;在事后解释模型中,回顾了推理模型的代表方法,并详细介绍提供事后解释的两类解释方法.此外,还总结了可解释知识推理在医疗、金融领域的应用.随后,对可解释知识推理的现状进行概述,最后展望了可解释知识推理的未来发展方向,以期进一步推动可解释推理的发展和应用.     

小样本关系下的时序知识图谱推理

在各大知识推理应用场景下,知识图谱中时序的缺失、知识图谱构建时实体关系的不完善,已然成为研究者们亟需解决的问题。为此,构造了一种融合时序信息与小样本关系的知识图谱推理模型,该模型将知识图谱的三元组表示扩展到含有时序信息的四元组表示,并通过时序信息来提高推理路径的准确性。此外,通过元学习从高频关系中学习元参数,并使用元参数适配小样本关系任务,提高模型在小样本关系中的泛化能力。实验表明,所提出的方法在Hits@1、Hits@3、Hits@10和MRR上均高于对比方法,并且均提高5%以上,表明所提出的方法可以完成知识图谱的推理,并且在小样本关系下具有较好的效果。     

面向时间感知的知识超图链接预测

知识超图是一种使用多元关系表示现实世界的异构图,但无论在通用领域还是垂直领域,现有的知识超图普遍存在不完整的情况.因此,如何通过知识超图中已有的链接推理缺失的链接是一个具有挑战性的问题.目前大多数研究使用基于多元关系的知识表示学习方法完成知识超图的链接预测任务,但这些方法仅从时间未知的超边中学习实体与关系的嵌入向量,没有考虑时间因素对事实动态演变的影响,导致在动态环境中的预测性能较差.首先,根据本文首次提出的时序知识超图定义,提出时序知识超图链接预测模型,同时从实体角色、位置和时序超边的时间戳中学习实体的静态表征和动态表征,以一定比例融合后作为实体嵌入向量用于链接预测任务,实现对超边时序信息的充分利用.同时,从理论上证明模型具有完全表达性和线性空间复杂度.此外,通过上市公司的公开经营数据构建时序知识超图数据集CB67,并在该数据集上进行了大量实验评估.实验结果表明:模型能够在时序知识超图数据集上有效地执行链接预测任务.     

时序知识图谱表示学习

知识图谱作为一种结构化的人类知识形式,对海量多源异构异质的数据语义互通起到了很好的支撑作用,并有效地支持了数据分析等任务,成为了近年来学术界和工业界的研究热点。目前大多数知识图谱都是根据非实时的静态数据构建,没有考虑实体和关系的时间特性。然而社交网络通信、金融贸易、疫情传播网络等应用场景的数据具有实时动态的特点以及复杂的时间特性,如何利用时序数据构建知识图谱并且对该知识图谱进行有效建模是一个具有挑战性的问题。目前,有许多研究工作利用时序数据中的时间信息丰富知识图谱的特征,赋予知识图谱动态特征,将事实三元组拓展为(头实体,关系,尾实体,时间)的四元组表示,使用时间相关四元组进行知识表示的知识图谱被统称为时序知识图谱。文中对时序知识图谱相关文献进行整理和分析,并对时序知识图谱表示学习的工作进行了全面综述。具体地,首先简单介绍了时序知识图谱的背景与定义;其次总结了时序知识图谱表示学习方法相比传统知识图谱表示学习方法的优点;然后从事实的建模方法角度详细阐述了时序知识图谱表示学习的主要方法,并且介绍了上述方法使用到的数据集;最后对该技术的主要挑战进行了总结,并对其未来研究方向进行了展望。     

一种面向临床领域时序知识图谱的链接预测模型

知识图谱(knowledge graph)链接预测可以解决知识图谱中缺失信息的发现和还原,是目前知识图谱领域的研究热点.传统的知识图谱链接预测方法大多面向静态的数据,并不适用于具有动态变化特性的时序知识图谱.时序知识图谱广泛存在于不同领域中,以临床医学领域为例,糖尿病作为一种典型的慢性病,其病程是一个疾病缓慢发展演化的过程.因此,在临床医学时序知识图谱上进行临床意义的链接预测,比如预测糖尿病的并发症,则需要考虑糖尿病病程发展随时间变化的时序特性,这也为传统的知识图谱链接预测方法带来巨大挑战.为此,结合临床医学事实知识的时序特性,提出一种基于LSTM序列增量学习的临床领域时序知识图谱链接预测模型.该模型结合LSTM长短期记忆单元递归神经网络在序列学习上的优势,通过构建基于LSTM的序列增量学习层,以端到端的方式提取时序知识图谱中的三元组时序特征,从而实现对时序知识图谱的链接预测.通过在糖尿病时序知识图谱上的实验,验证了模型的高效性、可用性及稳定性.     

基于表示学习和语义要素感知的关系推理算法

基于知识表示的关系推理方法研究是近年来统计关系学习和知识图谱领域共同关注的热点.通过对当前流行的基于知识表示的推理模型进行比较,分析了现有模型所普遍采用的基本假设存在的不合理之处,即忽视了实体与关系在语义上的多样性.据此提出了一种新的关系推理建模假设：实体对之间的每种关系反映的是两侧实体在某些特定方面的语义关联,通过对实体向量的语义方面要素进行选择性加权,可以实现对不同关系语义的表示和区分.根据该假设提出了一种新的关系推理建模方法,采用非线性变换的方法来解决表示学习中的语义分辨率问题.在公开数据集上的实验结果表明：所提出的算法对复杂关系类型和相关实体具有良好的语义区分能力,能有效提高知识图谱上的关系推理准确率,性能显著优于目前主流的相关工作.     

FS-Net: Frequency Statistical Network for Temporal Knowledge Graph Reasoning

Temporal Knowledge Graph (TKG) reasoning has attracted wide attention of researchers. Existing TKG reasoning methods have made great progress through modeling historical information. However, the time variability and unseen entities (relations) are still two major challenges that hinder the further improvement of this field. Moreover, since the structural information and temporal dependencies of the historical subgraph sequence have to be modeled, the traditional embedding-based methods often have high time consumption in the training and predicting processes, which greatly limits the application of the reasoning model in real-world scenarios. To address these issues, in this paper we propose a frequency statistical network for TKG reasoning, namely FS-Net. On the one hand, FS-Net continuously generates time-varying scores for the predictions at the changing timestamps based on the latest short-term historical fact frequency statistics; on the other hand, based on the fact frequency statistics at the current timestamp, FS-Net supplements the historical unseen entities (relations) for the predictions; in particular, FS-Net does not need training and has a very high time efficiency. Plenty of experiments on two TKG benchmark datasets demonstrate that FS-Net outperforms the baseline models.     

Uncertain knowledge graph embedding: an effective method combining multi-relation and multi-path

Uncertain Knowledge Graphs (UKGs) are used to characterize the inherent uncertainty of knowledge and have a richer semantic structure than deterministic knowledge graphs. The research on the embedding of UKG has only recently begun, Uncertain Knowledge Graph Embedding (UKGE) model has a certain effect on solving this problem. However, there are still unresolved issues. On the one hand, when reasoning the confidence of unseen relation facts, the introduced probabilistic soft logic cannot be used to combine multi-path and multi-step global information, leading to information loss. On the other hand, the existing UKG embedding model can only model symmetric relation facts, but the embedding problem of asymmetric relation facts has not be addressed. To address the above issues, a Multiplex Uncertain Knowledge Graph Embedding (MUKGE) model is proposed in this paper. First, to combine multiple information and achieve more accurate results in confidence reasoning, the Uncertain ResourceRank (URR) reasoning algorithm is introduced. Second, the asymmetry in the UKG is defined. To embed asymmetric relation facts of UKG, a multi-relation embedding model is proposed. Finally, experiments are carried out on different datasets via 4 tasks to verify the effectiveness of MUKGE. The results of experiments demonstrate that MUKGE can obtain better overall performance than the baselines, and it helps advance the research on UKG embedding.     

一种基于融合重构的子空间学习的零样本图像分类方法

图像分类是计算机视觉中一个重要的研究子领域.传统的图像分类只能对训练集中出现过的类别样本进行分类.然而现实应用中,新的类别不断涌现,因而需要收集大量新类别带标记的数据,并重新训练分类器.与传统的图像分类方法不同,零样本图像分类能够对训练过程中没有见过的类别的样本进行识别,近年来受到了广泛的关注.零样本图像分类通过语义空间建立起已见类别和未见类别之间的关系,实现知识的迁移,进而完成对训练过程中没有见过的类别样本进行分类.现有的零样本图像分类方法主要是根据已见类别的视觉特征和语义特征,学习从视觉空间到语义空间的映射函数,然后利用学习好的映射函数,将未见类别的视觉特征映射到语义空间,最后在语义空间中用最近邻的方法实现对未见类别的分类.但是由于已见类和未见类的类别差异,以及图像的分布不同,从而容易导致域偏移问题.同时直接学习图像视觉空间到语义空间的映射会导致信息损失问题.为解决零样本图像分类知识迁移过程中的信息损失以及域偏移的问题,本文提出了一种图像分类中基于子空间学习和重构的零样本分类方法.该方法在零样本训练学习阶段,充分利用未见类别已知的信息,来减少域偏移,首先将语义空间中的已见类别和未见类别之间的关系迁移到视觉空间中,学习获得未见类别视觉特征原型.然后根据包含已见类别和未见类别在内的所有类别的视觉特征原型所在的视觉空间和语义特征原型所在的语义空间,学习获得一个潜在类别原型特征空间,并在该潜在子空间中对齐视觉特征和语义特征,使得所有类别在潜在子空间中的表示既包含视觉空间下的可分辨性信息,又包含语义空间下的类别关系信息,同时在子空间的学习过程中利用重构约束,减少信息损失,同时也缓解了域偏移问题.最后零样本分类识别阶段,在不同的空间下根据最近邻算法对未见类别样本图像进行分类.本文的主要贡献在于:一是通过对语义空间中类别间关系的迁移,学习获得视觉空间中未见类别的类别原型,使得在训练过程中充分利用未见类别的信息,一定程度上缓解域偏移问题.二是通过学习一个共享的潜在子空间,该子空间既包含了图像视觉空间中丰富的判别性信息,也包含了语义空间中的类别间关系信息,同时在子空间学习过程中,通过重构,缓解知识迁移过程中信息损失的问题.本文在四个公开的零样本分类数据集上进行对比实验,实验结果表明本文提出的零样本分类方法取得了较高的分类平均准确率,证明了本文方法的有效性.     

QUBICS: A general quantity base management system

The problem of ambiguities has presented an obstacle to the practical use of qualitative simulation in dealing with real-life dynamic systems. One of the dominant causes of these prevailing ambiguities is the parsimonious use of information by current qualitative simulation models. Humans seem to utilize more quantity information to produce less ambiguous predictions when reasoning on dynamic mechanisms. To cover such human algebraic reasoning, the representation and processing of quantity information should be extended. This paper presents an inference system specialized for processing quantity information in order to support qualitative simulation. The system stores the information in the form of constraints into a quantity base, a notion that is parallel to the symbolic fact base, and processes queries based on linear programming techniques and goal-tree search. Temporal reasoning is also facilitated through an improved utilization of derivative information.     

利用深度卷积神经网络提高未知噪声下的语音增强性能

为了进一步提高基于深度学习的语音增强方法在未知噪声下的性能,本文从神经网络的结构出发展开研究.基于在时间与频率两个维度上,语音和噪声信号的局部特征都具有强相关性的特点,采用深度卷积神经网络（Deep convolutional neural network,DCNN）建模来表示含噪语音和纯净语音之间的复杂非线性关系.通过设计有效的训练特征和训练目标,并建立合理的网络结构,提出了基于深度卷积神经网络的语音增强方法.实验结果表明,在未知噪声条件下,本文方法相比基于深度神经网络（Deep neural network,DNN）的方法在语音质量和可懂度两种指标上都有明显提高.     

面向GAN生成图像的被动取证及反取证技术综述

生成对抗网络（generative adversarial network,GAN）快速发展,并在图像生成和图像编辑技术等多个方面取得成功应用。然而,若将上述技术用于伪造身份或制作虚假新闻,则会造成严重的安全隐患。多媒体取证领域的研究者面向GAN生成图像已提出了多种被动取证与反取证方法,但现阶段缺乏相关系统性综述。针对上述问题,本文首先阐述本领域的研究背景和研究意义,然后分析自然图像采集与GAN图像生成过程的区别。根据上述理论基础,详细介绍了现有GAN生成图像的被动取证技术,包括：GAN生成图像检测算法,GAN模型溯源算法和其他相关取证问题。此外,针对不同应用场景介绍基于GAN的反取证技术。最后,通过实验分析当前GAN生成图像被动取证技术所面临的挑战。本文根据对现有技术从理论和实验两方面的分析得到以下结论：现阶段,GAN生成图像的被动取证技术已在空间域和频率域形成了不同技术路线,较好地解决了简单场景下的相关取证问题。针对常见取证痕迹,基于GAN的反取证技术已能够进行有效隐藏。然而,该领域研究仍存在诸多局限：1）取证与反取证技术的可解释性不足;2）取证技术鲁棒性和泛化性较弱;3）反取证技术缺乏多特征域协同的抗分析能力等。上述问题和挑战还需要研究人员继续深入探索。     

Control loop diagnosis with ambiguous historical operating modes: Part 1. A proportional parametrization approach

Control-loop performance assessment methods have been evolving over the past two decades, with many different monitor algorithms being used to single out specific problems and determine the operating mode. However, a change in operating mode may affect multiple monitors, resulting in the possibility of conflicting assessments. Data-driven Bayesian methods were previously proposed which use multiple monitors to yield probabilistic assessments; however, training data for Bayesian methods requires complete knowledge of underlying operational modes. This paper proposes an approach based on proportionality parameters θ to address the problem of incomplete mode information in the training data; values in θ can be used to fill in missing information, and by varying θ one can determine the boundaries on a probabilistic diagnosis. Two diagnostic approaches are considered: the first type is direct probability approach, which can only be applied when historical data on the operation mode is sufficient and representative. The second type is the likelihood approach which can be applied to more general cases, including when the historical data is too limited to adequately represent mode frequency. In order to represent mode frequency, the likelihood approach takes into account prior probabilities of operating modes. The proposed methods are evaluated in two simulated chemical processes.     

Heterogeneous relational reasoning in knowledge graphs with reinforcement learning

Path-based relational reasoning over knowledge graphs has become increasingly popular due to a variety of downstream applications such as question answering in dialogue systems, fact prediction, and recommendation systems. In recent years, reinforcement learning (RL) based solutions for knowledge graphs have been demonstrated to be more interpretable and explainable than other deep learning models. However, the current solutions still struggle with performance issues due to incomplete state representations and large action spaces for the RL agent. We address these problems by developing HRRL (Heterogeneous Relational reasoning with Reinforcement Learning), a type-enhanced RL agent that utilizes the local heterogeneous neighborhood information for efficient path-based reasoning over knowledge graphs. HRRL improves the state representation using a graph neural network (GNN) for encoding the neighborhood information and utilizes entity type information for pruning the action space. Extensive experiments on real-world datasets show that HRRL outperforms state-of-the-art RL methods and discovers more novel paths during the training procedure, demonstrating the explorative power of our method.     

基于BERT和路径对比学习的归纳关系预测

在以往的知识图谱关系预测任务中,主要方法仅限于直推式推理,它们在新出现实体和关系情况下不能利用先验知识去处理归纳学习的问题。提出了基于BERT与路径对比学习的关系预测方法（BERT-based and path comparison learning,BPCL）。首先,利用卷积神经网络捕获子图目标三元组的上下文邻域信息,并将子图线性化为关系路径,利用BERT初始化边特征;其次,引入正、负关系路径;最后,联合对比学习和自监督学习训练对新出现实体之间的关系预测。在适用于归纳推理方法的常用基准数据集上,验证了该模型的预测精度有所提高。     

Future Event Prediction Based on Temporal Knowledge Graph Embedding

Accurate prediction of future events brings great benefits and reduces losses for society in many domains, such as civil unrest, pandemics, and crimes. Knowledge graph is a general language for describing and modeling complex systems. Different types of events continually occur, which are often related to historical and concurrent events. In this paper, we formalize the future event prediction as a temporal knowledge graph reasoning problem. Most existing studies either conduct reasoning on static knowledge graphs or assume knowledges graphs of all timestamps are available during the training process. As a result, they cannot effectively reason over temporal knowledge graphs and predict events happening in the future. To address this problem, some recent works learn to infer future events based on historical event-based temporal knowledge graphs. However, these methods do not comprehensively consider the latent patterns and influences behind historical events and concurrent events simultaneously. This paper proposes a new graph representation learning model, namely Recurrent Event Graph ATtention Network (RE-GAT), based on a novel historical and concurrent events attention-aware mechanism by modeling the event knowledge graph sequence recurrently. More specifically, our RE-GAT uses an attention-based historical events embedding module to encode past events, and employs an attention-based concurrent events embedding module to model the associations of events at the same timestamp. A translation-based decoder module and a learning objective are developed to optimize the embeddings of entities and relations. We evaluate our proposed method on four benchmark datasets. Extensive experimental results demonstrate the superiority of our RE-GAT model comparing to various baselines, which proves that our method can more accurately predict what events are going to happen.