检索结果-内蒙古大学图书馆

arXiv 2017年

作者： Liang, Bin Li, Hongcheng Su, Miaoqiang Bian, Pan Li, Xirong Shi, Wenchang School of Information Renmin University of China Beijing China Key laboratory of Data Engineering and Knowledge Engineering MOE Beijing China

In this paper, we present an effective method to craft text adversarial samples, revealing one important yet underestimated fact that DNN-based text classifiers are also prone to adversarial sample attack. Specifically, confronted with different adversarial scenarios, the text items that are important for classification are identified by computing the cost gradients of the input (white-box attack) or generating a series of occluded test samples (black-box attack). Based on these items, we design three perturbation strategies, namely insertion, modification, and removal, to generate adversarial samples. The experiment results show that the adversarial samples generated by our method can successfully fool both state-of-the-art character-level and word-level DNN-based text classifiers. The adversarial samples can be perturbed to any desirable classes without compromising their utilities. At the same time, the introduced perturbation is difficult to be perceived. Copyright © 2017, The Authors. All rights reserved.

关键词： Classification (of information)

来源：评论

学校读者我要写书评

暂无评论

Detecting the research diversity of researchers in library and information science:An exploratory study

引用

data Science and Informetrics 2022年第3期2卷 22-36页

作者： Yuehua Zhao Sicheng Zhu Jie Wu Hao Wang Sanhong Deng Jiangsu Key Laboratory of Data Engineering and Knowledge Service Nanjing UniversityNanjingChina School of Information Management Nanjing UniversityNanjingChina

Metrics have emerged as an important tool for quantitatively evaluating researchers from a variety of perspectives,including research impact,research quality,interdisciplinarity,and *** in the field of library and information science,many previous studies have highlighted the characteristics of researchers in this ***,only a minority of the studies address the aspect of diversity in research *** purpose of this study is to(1)evaluate the topic diversity of researchers in library and information science and(2)examine the relationships between the researcher topic diversity and research *** propose an indicator to quantify author topic diversity,which we refer to as author topic diversity(ATD).Latent Dirichlet Allocation(LDA)is used to detect topics in the field,while cosine similarity is used to calculate the diversity of research topics in a given researcher’s *** results show that topic diversity in the field of library and information science varies greatly from author to *** addition,weak positive correlations are found between the ATD and citation indicators,suggesting that engaging in diversified topics may lead to higher research impact.

关键词： Research diversity Research evaluation Library and information science

来源：评论

学校读者我要写书评

暂无评论

RAKA: Co-training of relationships and attributes for cross-lingual knowledge alignment

arXiv

引用

arXiv 2019年

作者： Chen, Bo Zhang, Jing Tang, Xiaobin Chen, Hong Li, Cuiping Key Laboratory of Data Engineering Knowledge Engineering of Ministry of Education Renmin University of China Information School Renmin University of China

Cross-lingual knowledge alignment suffers from the attribute heterogeneity when leveraging the attributes and also suffers from the conflicts when combing the results inferred from attributes and relationships. This paper proposes an interaction-based attribute model to capture the attribute-level interactions for estimating entity similarities, eliminating the negative impact of the dissimilar attributes. A matrix-based strategy is adopted in the model to accelerate the similarity estimation. We further propose a co-training framework together with three merge strategies to combine the alignments inferred by the attribute model and the relationship model. The whole framework can effectively and efficiently infer the aligned entities, relationships, attributes, and values simultaneously. Experimental results on several cross-lingual knowledge datasets show that our model significantly outperforms state-of-the-art comparison methods (improving 2.35-51.57% in terms of Hit Ratio@1). Copyright © 2019, The Authors. All rights reserved.

关键词： Alignment

来源：评论

学校读者我要写书评

暂无评论

Selectivity Estimation of Correlated Properties in RDF data for SPARQL Query Optimization

Selectivity Estimation of Correlated Properties in RDF Data ...

引用

Fifth International Conference on Semantics,knowledge and Grid(第五届语义、知识与网格国际会议 SKG 2009)

作者： Bin Lv Xiaoyong Du Yan Wang Key Laboratory of Data Engineering and Knowledge Engineering Ministry of Education School of Information Renmin University of China Beijing 100872 China

Nowadays mainstream RDF Repository Systems are based on RDBMS. The SPARQL query engine translates a SPARQL query into a SQL one, and then the RDBMS executes the SQL query. However the RDBMS optimizers, which usually assume that columns are statistically independent, often underestimate the selectivity of conjunctive predicates and choose a bad query execution plan. It is important for query optimizers to detect correlations among properties. We propose a way of computing property correlations based on ontology itself in order to improve the execution performance of the SQL translated from SPARQL statement queries.

关键词： Ontology SPARQL SQL Property Correlation Query optimization

来源：评论

学校读者我要写书评

暂无评论

Trust Service Discovery by Opinions Classification on Virtual Communities

Trust Service Discovery by Opinions Classification on Virtua...

引用

International Conference on Networks Security, Wireless Communications and Trusted Computing, NSWCTC

作者： Qing Zhu Department of Computer Science Information School Key Laboratory of Data Engineering and Knowledge Engineering Renmin University of China Beijing China

ISBN: (纸本)9781424465972;9780769540115

Internet has become an excellent ecommerce platform for bringing together large numbers of buyers and sellers across wide geographic regions. Trust and reputation systems represent a. significant trend in decision support for Internet mediated service provision. However, most existing work assumes that all users have the same trust metrics, but in real life different users often have different preference of product attributes. This paper proposes trusted query navigation model by analyzing online customer reviews of trustworthiness of websites. The first step analyzes counting history feedback in the system offline and generates dynamic trust model by opinion classification. The second step presents trust evaluation ranking to help the user can easily discovery trust service of matching his needs. The experimental evaluation shows that the trusted query evaluation ranking has high trading efficiency, quick learning ability and satisfactory performance.

关键词： Business Feedback Portals History Web services Computer security data security Information security Wireless communication Computer networks

来源：评论

学校读者我要写书评

暂无评论

Automatic Fusion for Multimodal Entity Alignment: A New Perspective from Automatic Architecture Search

Automatic Fusion for Multimodal Entity Alignment: A New Pers...

引用

IEEE International Conference on Multimedia and Expo (ICME)

作者： Chenyang Bu Yunpeng Hong Shiji Zang Guojie Chang Xindong Wu Key Laboratory of Knowledge Engineering With Big Data (the Ministry of Education of China) School of Computer Science and Engineering HeFei University of Technology

ISBN: (数字)9798350390155

ISBN: (纸本)9798350390162

Integrating multimodal data from diverse sources is crucial for enhancing various applications. Multimodal entity alignment (MMEA), which discovers equivalent entities across different sources and modalities, aims to eliminate data silos for comprehensive integration. A key challenge in MMEA is effectively fusing vector representations from different modalities of the same entity for optimal entity matching. Existing fusion methods involve individual fusion operators (e.g., concatenation and summation) or the manual design of complex network structures, incurring significant human resource costs. In this paper, for the first time, we introduce the research question of automatic fusion for MMEA and propose an efficient approach from the perspective of automated architecture search. Experimental comparisons with state-of-the-art methods on real-world datasets demonstrate the effectiveness of the proposed approach.

关键词： Costs Search methods Complex networks Vectors

来源：评论

学校读者我要写书评

暂无评论

Law Text Clustering Based on Referential Relations

Law Text Clustering Based on Referential Relations

引用

ChinaGrid Annual Conference (ChinaGrid)

作者： Biao Fan Tao Liu He Hu Xiaoyong Du Ministry of Education Key Laboratory of Data Engineering and Knowledge Engineering China School of Information Renmin University of China Beijing China

This paper proposes a new method to cluster law texts based on referential relation of laws. We extract law entities (an entity represents a law) and their referential relation from law texts. Then SimRank algorithm is applied to calculate law entity's similarity through referential relation and law clustering is carried out based on the SimRank similarity. This is the first time to apply SimRank algorithm in the domain of Law and use it to carry out text clustering. Prototype and experiments show that our solution is feasible. We also publish the extracted data as Linked Law data with RDF data model, which forms the first open semantic web database in Law domain. Linked Law data enables user to access law data with rich data links and query web data by application interface of Semantic Web.

关键词： Clustering algorithms Heuristic algorithms databases Algorithm design and analysis Resource description framework data mining Semantics

来源：评论

学校读者我要写书评

暂无评论

OLTP workloads on modern processor: Characterization and analysis

引用

Journal of Computational Information Systems 2008年第1期4卷 389-394页

作者： Liu, Dawei Qin, Biao Wang, Shan Gong, Weiwei School of Information Renmin University of China Beijing 100872 China Key Laboratory of Data Engineering and Knowledge Engineering Renmin University of China Beijing 200872 China

This paper analysis of how OLTP workloads interact with modern processors and caches behavior. First, we extend TPC-C, the OLTP-oriented benchmark, to ETPC-C benchmark, for measuring the performance of main-memory database (MMDBMS) more precisely. As the performance of MMDBMS is not affected by disk I/O, it is more sensitive to cache usage. Then using ETPC-C benchmark, we investigated the behavior of caches and processors extensively. We find that the miss stall time is mostly spent on on-CPU-chip caches, that is, the first and second level cache misses are dominant. Furthermore, we find instruction cache (I-cache) stall time of on-CPU-chip is a major component to the memory stall time. The smaller the emulated users, the more proportion the I-cache stall time of on-CPU-chip contributes to the memory stall time. However, if employing index, the system under test (SUT) has more total I-cache stall time than the SUT without index at the same number of emulated users and data population. Another observation is that the SUT with index has a little more branch misprediction rate than the SUT without index in average. Finally, we find only the third level (L3) D-cache stall time rate increases with the number of users. This is because L3 D-cache miss incremental rate is the largest. Under TPC-and ETPC-evaluation, we find that for optimized database performance on modern computers, reducing instruction miss penalty is equally important to reducing data miss penalty;since they are conflict efforts, the best way is to have them balanced.

关键词： Program processors

来源：评论

学校读者我要写书评

暂无评论

P2 Law: Scaling Law for Post-Training After Model Pruning

arXiv

引用

arXiv 2024年

作者： Chen, Xiaodong Hu, Yuxuan Zhang, Xiaokang Wang, Yanling Li, Cuiping Chen, Hong Zhang, Jing School of Information Renmin University of China Beijing China Key Laboratory of Data Engineering and Knowledge Engineering Beijing China Zhongguancun Laboratory China

Pruning has become a widely adopted technique for reducing the hardware requirements of large language models (LLMs). To recover model performance after pruning, post-training is commonly employed to mitigate the resulting performance degradation. While post-training benefits from larger datasets, once the dataset size is already substantial, increasing the training data provides only limited performance gains. To balance post-training cost and model performance, it is necessary to explore the optimal amount of post-training data. Through extensive experiments on the Llama-3 and Qwen-2.5 series models, pruned using various common pruning methods, we uncover the scaling Law for Post-training after model Pruning, referred to as the P2 Law. This law identifies four key factors for predicting the pruned model’s post-training loss: the model size before pruning, the number of post-training tokens, the pruning rate, and the model’s loss before pruning. Moreover, P2 Law can generalize to larger dataset sizes, larger model sizes, and higher pruning rates, offering valuable insights for the post-training of pruned LLMs. © 2024, CC BY.

关键词： Digital elevation model

来源：评论

学校读者我要写书评

暂无评论

Enabling Efficient NVM-Based Text Analytics without Decompression

Enabling Efficient NVM-Based Text Analytics without Decompre...

引用

International Conference on data engineering

作者： Xiaokun Fang Feng Zhang Junxiang Nong Mingxing Zhang Puyun Hu Yunpeng Chai Xiaoyong Du Key Laboratory of Data Engineering and Knowledge Engineering (MOE) and School of Information Renmin University of China Department of Computer Science and Engineering Tsinghua University

ISBN: (数字)9798350317152

ISBN: (纸本)9798350317169

Text analytics directly on compression (TADOC) is a promising technology designed for handling big data analytics. However, a substantial amount of DRAM is required for high performance, which limits its usage in many important scenarios where the capacity of DRAM is limited, such as memory-constrained systems. Non-volatile memory (NVM) is a novel storage technology that combines the advantage of reading per-formance and byte addressability of DRAM with the durability of traditional storage devices like SSD and HDD. Unfortunately, no research demonstrates how to use NVM to reduce DRAM utilization in compressed data analytics. In this paper, we propose N-TADOC, which substitutes DRAM with NVM while maintaining TADOC's analytics performance and space savings. Utilizing an NVM block device to reduce DRAM utilization presents two challenges, including poor data locality in traversing datasets and auxiliary data structure reconstruction on NVM. We develop novel designs to solve these challenges, including a pruning method with NVM pool management, bottom-up upper bound estimation, correspondent data structures, and persistence strategy at different levels of cost. Experimental results show that on four real-world datasets, N-TADOC achieves 2.04× performance speedup compared to the processing directly on the uncompressed data and 70.7% DRAM space saving compared to the original TADOC.

关键词： Performance evaluation Upper bound data analysis Costs Nonvolatile memory Random access memory Estimation

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：