检索结果-内蒙古大学图书馆

A review on the reliability of knowledge graph: from a knowledge representation learning perspective

WORLD WIDE WEB-INTERNET AND WEB INFORMATION SYSTEMS 2025年第1期28卷 1-38页

作者： Yang, Yunxiao Chen, Jianting Xiang, Yang Tongji Univ Coll Elect & Informat Engn Caoan Highway Shanghai 201804 Peoples R China

knowledge graphs manage and organize data and information in a structured form, which can provide effective support for various applications and services. Only reliable knowledge can provide valuable information. However, most existing knowledge graphs encounter the problem of partially unreliable knowledge. With the progress of the Internet and information technology, how to ensure the reliability of knowledge graphs has become a significant research topic. We first clarify the concept of knowledge graph reliability based on the attributes of facts in knowledge graphs. It includes two parts: the correctness and uncertainty of knowledge. We then analyze their corresponding research tasks. The research of knowledge correctness aims to handle the erroneous triples in knowledge graphs, whereas the research of knowledge uncertainty assesses the ambiguous and probabilistic triples. knowledge representation learning, a neural technique to process symbolic knowledge, is the promising approach in the research of knowledge graph reliability. Therefore, we summarize the related studies on knowledge correctness and uncertainty based on the framework of knowledge representation learning, which includes four categories: score function modification, representation vector optimization, loss function adjustment, and textual information integration. Additionally, we present an analysis of the widely used benchmarks, and lastly conclude with a discussion on the potential trends and future research directions in the reliability of knowledge graph.

关键词： knowledge graph knowledge reliability knowledge representation learning Uncertainty measurement Error detection

来源：评论

学校读者我要写书评

暂无评论

knowledge is power: Open-world knowledge representation learning for knowledge-based visual reasoning ☆,☆☆ , ☆☆

引用

ARTIFICIAL INTELLIGENCE 2024年 333卷

作者： Zheng, Wenbo Yan, Lan Wang, Fei-Yue Wuhan Univ Technol Sch Comp Sci & Artificial Intelligence Wuhan 430070 Peoples R China Hunan Univ Coll Comp Sci & Engn Changsha 410082 Peoples R China Chinese Acad Sci Inst Automat Beijing 100190 Peoples R China

knowledge-based visual reasoning requires the ability to associate outside knowledge that is not present in a given image for cross-modal visual understanding. Two deficiencies of the existing approaches are that (1) they only employ or construct elementary and explicit but superficial knowledge graphs while lacking complex and implicit but indispensable cross-modal knowledge for visual reasoning, and (2) they also cannot reason new/ unseen images or questions in open environments and are often violated in real-world applications. How to represent and leverage tacit multimodal knowledge for open-world visual reasoning scenarios has been less studied. In this paper, we propose a novel open-world knowledge representation learning method to not only construct implicit knowledge representations from the given images and their questions but also enable knowledge transfer from a known given scene to an unknown scene for answer prediction. Extensive experiments conducted on six benchmarks demonstrate the superiority of our approach over other state-of-the-art methods. We apply our approach to other visual reasoning tasks, and the experimental results show that our approach, with its good performance, can support related reasoning applications.

关键词： Visual reasoning knowledge representation learning Open-world learning Graph model

来源：评论

学校读者我要写书评

暂无评论

Investigating Domain knowledge Graph knowledge Reasoning and Assessing Quality Using knowledge representation learning and knowledge Reasoning Algorithms

引用

JOURNAL OF INFORMATION & knowledge MANAGEMENT 2025年第1期24卷

作者： Cao, Ying Ningbo Polytech Sch Artificial Intelligence Ningbo 315800 Zhejiang Peoples R China

Domain knowledge Graphs are becoming increasingly significant in a multitude of sectors, as they collate distinctive data from a plethora of disciplines. However, in some domains, generic knowledge representation is limited, resulting in problems like duplication and knowledge loss. To solve the above issues, the study first builds a knowledge representation learning model for training and then constructs a Domain knowledge Graph inference algorithm based on knowledge representation learning for knowledge inference and quality evaluation. The results indicated that the effectiveness of the knowledge-inference-raised method was the best in datasets of different sizes, with the average accuracy, Hits@1, Hits@3 and Hits@10 of 0.906, 0.914, 0.942 and 0.948, respectively. In the results of the link prediction task, the study method had a poor performance for only one relation, with an average accuracy rate of 0.675. In the application results of knowledge graph quality assessment, the interpretability and data fusion ability of both Domain knowledge Graphs were strong, with the accuracy indexes exceeding 0.95, and the various indexes of simplicity and completeness exceeding 0.88. The above results indicate that our research method is effective in filling in the missing knowledge through knowledge reasoning and ensures data reliability.

关键词： knowledge graph inference algorithm knowledge representation learning link prediction quality evaluation

来源：评论

学校读者我要写书评

暂无评论

Evolving knowledge representation learning with the dynamic asymmetric embedding model

引用

EVOLVING SYSTEMS 2024年第6期15卷 2323-2338页

作者： Khan, Muhib A. Khan, Saif Ur Rehman Haider, Syed Zohair Quain Khan, Shakeeb A. Bilal, Omair Univ Engn & Technol Dept Comp Sci Taxila Pakistan Cent South Univ Sch Comp Sci & Engn Changsha Peoples R China Inst Southern Punjab Dept Comp Sci & IT Multan Punjab Pakistan

Detecting errors in knowledge Graphs is challenging due to the scarcity of ground-truth labels and the unpredictable nature of error patterns, leading to noticeable noise affecting downstream tasks. KG's repositories contain billions of triplets representing relationships as instances (head entity, relation, tail entity). This distinctive and symbolic architecture facilitates advanced knowledge exploitation techniques aimed at improving the accuracy of other intelligence-driven applications. Making logical rules to validate triples is a traditional approach, yet it lacks generalizability due to the variability of rules across knowledge graphs stemming from domain-specific knowledge. Recent models such as TransE, TransH, TransR, and TransD have approached the task of knowledge graph completion by treating relationships as translations from head entities to tail entities. However, these models exhibit limitations in capturing the complexity and diversity of entities and relations within knowledge graphs, particularly with regards to symmetric, one-to-many, and many-to-many relations. To address these limitations, this paper introduces a novel model called TransDAE for dynamic asymmetrical embedding in knowledge graph completion, building upon the foundation of TransDR. Unlike TransDR, which focuses less on asymmetrical relations and overlooks the imbalanced characteristics of relationships, such as one-to-many and many-to-one, TransDAE considers the distinct properties of head and tail entities in similar relations. Specifically, TransDAE incorporates two vectors for each triple relationship, weighting each entity vector by its corresponding head and tail relation vectors in the relation embedding space. This enhances the model's flexibility and improves its ability to interpret and uncover latent attributes of entities and relations. Through experimental evaluation on tasks including triplet classification and link prediction, TransDAE demonstrates superior performan

关键词： knowledge representation learning Embedding model NLP knowledge graph

来源：评论

学校读者我要写书评

暂无评论

A joint knowledge representation learning of sentence vectors weighting and primary neighbor constraints

引用

knowledge AND INFORMATION SYSTEMS 2024年第11期66卷 6739-6760页

作者： Zhao, Erping Chen, Bailin BianBaDroMa Ngodrup Xizang Minzu Univ Coll Informat Engn Xianyang 712082 Shaanxi Peoples R China Tibet Net Cloud Scitech Co Ltd Lhasa 850001 Xizang Peoples R China

knowledge representation learning is to project entities and relations into low-dimensional vector spaces, and efficiently calculate entities, relations, and complex semantic associations between them. However, the translation models that fuse multi-source information fail to specially pay attention to the semantic information of important sentences in the textual descriptions of an entity, which lead to the low quality of embedded representation of textual descriptions, and fail to utilize the primary neighbor of the entity to further constrain the entity representation, resulting in the weak knowledge representation learning ability of the models. To address these issues, we propose a joint knowledge representation learning model of sentence vectors weighting and primary neighbor constraints called SW &NC-Joint. First, the weight distribution of the sentences is calculated according to the frequency of the neighbors in each sentence, and then the textual descriptions are represented by the sentence vectors weighting. Secondly, the concept of neighbor association degree is put forward, and the primary neighbor is filtered out for each entity according to the association degree. Finally, the weighted textual description information and triple structure information are jointly trained, the most crucial thing that the model is constrained by using the distance between the entity and its primary neighbor in the training process, so as to improve the representation learning ability of the model. The link prediction and triple classification experiments are carried out on FB15K and WN18 datasets. The experimental results demonstrate that our model significantly outperforms existing state-of-the-art text-enhanced knowledge representation learning model for link prediction and triple classification tasks.

关键词： knowledge graph knowledge representation learning Neighbor entity Sentence vector weighting Primary neighbor constraint

来源：评论

学校读者我要写书评

暂无评论

Large Language Model Enhanced knowledge representation learning: A Survey

引用

DATA SCIENCE AND ENGINEERING 2025年 1-24页

作者： Wang, Xin Chen, Zirui Wang, Haofen Hou, U. Leong Li, Zhao Guo, Wenbin Tianjin Univ Coll Intelligence & Comp Tianjin Peoples R China Tongji Univ Coll Design & Innovat Shanghai Peoples R China Univ Macao Fac Sci & Technol Macau Peoples R China

knowledge representation learning (KRL) is crucial for enabling applications of symbolic knowledge from knowledge Graphs (KGs) to downstream tasks by projecting knowledge facts into vector spaces. Despite their effectiveness in modeling KG structural information, KRL methods are suffering from the sparseness of KGs. The rise of Large Language Models (LLMs) built on the Transformer architecture presents promising opportunities for enhancing KRL by incorporating textual information to address information sparsity in KGs. LLM-enhanced KRL methods, including three key approaches, encoder-based methods that leverage detailed contextual information, encoder-decoder-based methods that utilize a unified Seq2Seq model for comprehensive encoding and decoding, and decoder-based methods that utilize extensive knowledge from large corpora, have significantly advanced the effectiveness and generalization of KRL in addressing a wide range of downstream tasks. This work provides a broad overview of downstream tasks while simultaneously identifying emerging research directions in these evolving domains.

关键词： knowledge graph Large language model knowledge representation learning

来源：评论

学校读者我要写书评

暂无评论

A knowledge representation learning MODEL BASED ON RELATION ROTATION IN TWO-DIMENSIONAL MINKOWSKI SPACE

引用

MATHEMATICAL FOUNDATIONS OF COMPUTING 2024年第4期7卷 429-438页

作者： Wang, M. I. N. G. J. I. E. Zhou, J. U. X. I. A. N. G. Wang, J. U. N. Gan, J. I. A. N. H. O. U. LI, Z. I. J. I. E. Yunnan Normal Univ Key Lab Educ Informatizat Nationalities Minist Educ Kunming 650500 Peoples R China Yunnan Normal Univ Yunnan Key Lab Smart Educ Kunming 650500 Peoples R China

The knowledge graph is a structured representation of knowledge, which typically employs a triplet to convey information. However, the long tail distribution that characterizes most large-scale knowledge bases poses serious challenges regarding entity and relationship information in the long-tail segment, such as data sparsity and incomplete knowledge. To address these challenges and enhance the completeness of the knowledge graph, this paper proposes a novel model for knowledge representation learning in the context of knowledge graph completion. Specifically, the model maps entities to a two-dimensional Minkowski space and defines relationships as rotations. Additionally, the Adam optimizer is employed to optimize the training process. Finally, the model's efficacy is evaluated via link prediction tasks using public datasets FB15k, WN18, FB15k-237, and WB18RR, with results being compared to those derived from classical knowledge representation models. The empirical analysis reveals that the proposed model achieves superior performance across most evaluation metrics.

关键词： knowledge graph completion knowledge representation learning 2D Minkowski space

来源：评论

学校读者我要写书评

暂无评论

KRL_Match: knowledge graph objects matching for knowledge representation learning

引用

knowledge AND INFORMATION SYSTEMS 2023年第2期65卷 641-681页

作者： Suo, Xinhua Guo, Bing Shen, Yan Dai, Shengxin Wang, Wei Chen, Yaosen Zhang, Zhen SiChuan Univ Sch Comp Sci & Engn 24South Sect 1Yihuan Rd Chengdu 610065 Sichuan Peoples R China Chengdu Univ Informat Technol Sch Comp Sci 24Sect 1XueFu Rd Chengdu 610225 Sichuan Peoples R China

The way of obtaining the embeddings of the knowledge graph objects through modeling with binary classification method from the level of triple structure is coarser in granularity for the existing knowledge representation learning models based on the probability, and the space-time efficiency of negative sampling is lower for the most of the knowledge representation learning models at present. To solve these problems, this paper proposes a knowledge representation learning model KRL_Match, which carries out the knowledge graph objects matching centered on a certain kind of knowledge graph objects (head entity, tail entity, relation), and executes multi-classification learning to determine the true matching and dynamic implicit negative sampling. Specifically, first, we make two classes of the knowledge graph objects of target and source in the same kind of knowledge graph objects matched mutually by their matrix multiplication operation in a knowledge graph batch sample space, which is constructed by random sampling from the universe set of the knowledge graph instance, and the knowledge graph objects matching sample spaces will be implicitly generated meanwhile;then, we measure the matching degree of each matching of the knowledge graph objects by softmax regression multi-classification method in each implicit sample space;finally, we fit the real probability with the matching degree by optimizing the cross-entropy loss based on the local closed world assumption. We conduct the knowledge graph objects matching for the knowledge representation learning inspired by the attention mechanism and firstly create the dynamic implicit negative sampling method in the knowledge representation learning. Experiments show that the KRL_Match model has achieved better performances compared with the baselines: Hits@10 (filter) has increased by 12.2% and 6.1% on benchmarks FB15K and FB15K237 respectively for the entity prediction task, and accuracy has increased by 12.6% on benchmark

关键词： knowledge graph objects matching Attention mechanism Multi-classification Dynamic implicit negative sampling knowledge representation learning knowledge graph embedding knowledge graph

来源：评论

学校读者我要写书评

暂无评论

Optimal Recommendation Models Based on knowledge representation learning and Graph Attention Networks

引用

IEEE ACCESS 2023年 11卷 19809-19818页

作者： He, Qing Liu, Songyan Liu, Yao Heilongjiang Univ Sch Elect Engn Harbin 150080 Heilongjiang Peoples R China

knowledge representation learning techniques process the knowledge graph, embedding entities and relationships into a continuous dense low-dimensional vector space, and providing the rich semantic association information between entities embedded in the knowledge graph to the recommendation module, improving recommendation performance and providing better interpretability. However, the knowledge representation learning techniques based on the translation model TransD with too many parameters and no association between entity representations are difficult to apply to large knowledge graph, as well as the problem is that most existing knowledge graph which based on recommendation systems ignore the different levels of importance that users attach to different relationships of items. To address these shortcomings, we propose an improved knowledge representation learning model Cluster TransD and a recommendation model Cluster TransD-GAT based on knowledge graph and graph attention networks, where the Cluster TransD model reduces the number of entity projections, makes the association between entity representations, reduces the computational pressure, and makes it better to be applied to the large knowledge graph, and the Cluster TransD-GAT model can capture the attention of different users to different relationships of items. Extensive comparison and ablation experiments on three real datasets show that the model proposed in this paper has significant performance improvements in terms of average ranking, accuracy and recall of the scoring function compared to other state-of-the-art models.

关键词： representation learning Recommender systems knowledge graphs learning systems knowledge representation knowledge graph knowledge representation learning graph attention networks personalized recommendation

来源：评论

学校读者我要写书评

暂无评论

Numerical knowledge representation learning and Link Prediction over knowledge Graph 20th

Numerical Knowledge Representation Learning and Link Predict...

引用

20th International Conference on Intelligent Computing (ICIC)

作者： Huang, Zhen Qiu, Xue Liu, Yu Dalian Univ Technol Dalian Liaoning Peoples R China

ISBN: (纸本)9789819756179;9789819756186

Link prediction has been extensively studied in knowledge graph (KG) completion tasks, whereas numerical attribute relation prediction studies are still relatively scarce due to the non-discrete challenges. Numerical attribute relations are pervasive in real-world KGs (e.g., relations in product graphs such as "car, has MaxPower, 103"), which imbue entities with rich semantics. This paper presents a novel framework for numerical relation prediction. The framework comprises a numerical relation knowledge representation method (NumKR) and a numerical relation prediction framework based on GEN. The aim is to enhance the accuracy of the numerical link prediction model in a realworld product knowledge graph. The experimental results demonstrate that our model outperforms the simple KGE model in terms of link prediction performance.

关键词： knowledge representation learning knowledge graph Numeric relation prediction

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：