检索结果-内蒙古大学图书馆

Knowledge graph based question-answering model with subgraph retrieval optimization

COMPUTERS & OPERATIONS RESEARCH 2025年 177卷

作者： Zhu, Rui Liu, Bo Tian, Qiuyu zhang, ruwen zhang, Shengxiang Hu, Yanna Cao, Jiuxin Southeast Univ Sch Cyber Sci & Engn Nanjing 211189 Peoples R China Southeast Univ Sch Comp Sci & Engn Nanjing 211189 Peoples R China Purple Mt Labs Nanjing 211111 Peoples R China

Knowledge graph-based question answering (QA) is a critical domain within natural language processing, aimed at delivering precise and efficient responses to user queries. Current research predominantly focuses on minimizing subgraph sizes to enhance the efficiency and compactness of the search space. However, natural language queries often exhibit ambiguities, and merely reducing subgraph sizes may overlook relevant answer entities. Additionally, redundant relationships among entities in the knowledge graph can adversely affect QA model performance. To address these limitations, this paper introduces a novel QA model that optimizes subgraph retrieval. The proposed model enhances entity linking and subgraph retrieval by leveraging contextual features from both questions and entities. It disambiguates entities using relevant contextual features and refines the search process through entity relation merging and entity ranking strategies. This methodology improves entity recognition and linking, reduces subgraph dimensions, and broadens answer coverage, resulting in substantial improvements in QA performance. Experimental results on the CCKS2019CKBQA dataset demonstrate the model & sacute;effectiveness, showing an average F1 score improvement of 2.99% over the leading baseline model. Furthermore, the model's application in the field of ocean engineering underscores its practical utility and significance.

关键词： Subgraph retrieval Entity disambiguation Intelligent question-answering

来源：评论

学校读者我要写书评

暂无评论

Automatic stomatal phenotyping of lettuce leaves for plant factory: An improved U-network approach

引用

INDUSTRIAL CROPS AND PRODUCTS 2025年 226卷

作者： zhang, Xihai Zhu, Jiaxi Cheng, Jin zhang, ruwen Xia, Juheng Guo, Ruichao Wang, Hao Xu, Yonghua Northeast Agr Univ Sch Elect & Informat Harbin 150030 Peoples R China Northeast Agr Univ Natl Key Lab Smart Farm Technol & Syst Harbin 150030 Peoples R China

Lettuce, a vegetable rich in nutritional and medicinal value, is commonly analyzed as a modern industrial crop through macroscopic factors such as temperature, humidity, and light, but its microscopic characteristics are still underexplored. At the microscopic scale, stomatal characteristics are the most indicative of lettuce growth status and serve as crucial pathways for plant gas exchange and carbon-water cycle regulation. Therefore, stomatal research is an important area in crop breeding and stress analysis, and stomatal feature detection is a key step in this field. Current traditional methods for stomatal feature measurement are inefficient, imprecise, and laborintensive. This study proposes a method for stomatal feature extraction of lettuce leaves based on an improved U-net network to improve measurement efficiency and accuracy. To this end, a dual symmetric path structure was designed, incorporating two independent decoding paths to separately extract global contextual information and local detail features, effectively integrating multi-scale information during the decoding phase via feature concatenation and convolutional fusion modules. To mitigate edge information loss caused by repeated down-sampling in the U-Net network, a hybrid dilated convolution module was incorporated into the encoding phase, with overlapping pooling replacing standard pooling to enhance the network's precision in recognizing small objects. Furthermore, the network incorporates a CBAM attention mechanism module to strengthen its capacity for extracting effective features. To optimize network performance, the NAdam optimization function was employed to speed up convergence and minimize computational resource consumption. The MFe (Measurement Feature) visualization and interaction system developed using OpenCV enables precise measurement of stomatal major and minor axes, area, and density for lettuce leaves. Experimental results indicate that the improved U-Net network achieved enha

关键词： Stomata Deep learning U -net CBAM Feature extraction

来源：评论

学校读者我要写书评

暂无评论

A crystallinity control strategy for controllable Heterogeneous-Homogeneous reaction and catalytic performance in wet-mechanochemical synthesized FeS2 Fenton system

引用

SEPARATION AND PURIFICATION TECHNOLOGY 2025年 355卷

作者： Wang, Mingzhe Shi, Xiaoguo Wen, Zunqing Gu, Yawei Feng, Qing Ma, Ke Chen, Qiucheng Zhu, Yiming zhang, ruwen zhang, Longquan zhang, Minglu Ren, Qian zhang, Rui Xu, Chongqing Qilu Univ Technol Shandong Acad Sci Sch Environm Sci & Engn Jinan 250353 Peoples R China Chinese Acad Sci Inst Met Res Shenyang Natl Lab Mat Sci Shenyang 110016 Peoples R China Qilu Univ Technol Ecol Inst Shandong Acad Sci Jinan 250013 Peoples R China

In heterogeneous Fenton system, the Heterogeneous-Homogeneous reaction type induced by surface-bonded or dissolved catalytic center plays a vital role in determining Fenton performance. However, how to directionally control the reaction type with facile method still remains challenging. Herein, a novel crystallinity control strategy based on wet-mechanochemical synthesis combined with heat-treatment was applied to synthesize FeS2 nanoparticles with different crystallinity. The different crystalline FeS2 based Fenton system exhibited superior sulfamethoxazole removal performance at wide pH range (5.6-11) with center dot OH as dominating reactive oxygen species. The relative degradation contribution of homogeneous and heterogeneous Fenton reaction was semi-quantitatively evaluated, and the relationship between crystallinity indexes of FeS2 and degradation kinetic constants was established. It's found that FeS2 crystallinity was directly correlated with its degradation efficiency, which first increased and then declined with crystallinity increasing. Crystallinity control strategy effectively determined dominating reaction type (homogeneous or heterogeneous) and further regulated catalytic performance in FeS2 Fenton system based on the interaction of specific surface area and Fe2+ dissolution ability. This work provides a novel crystallinity control strategy to build a correlative bridge among crystallinity, dominating reaction type and catalytic performance in FeS2 Fenton system.

关键词： FeS2 Fenton reaction Crystallinity Dominating reaction type Catalytic performance

来源：评论

学校读者我要写书评

暂无评论

METER: Multimodal Hallucination Detection with Mixture of Experts via Tools Ensembling and Reasoning 13th

METER: Multimodal Hallucination Detection with Mixture of Ex...

引用

13th International Conference on Natural Language Processing and Chinese Computing

作者： zhang, ruwen Chen, Jinglu Dai, Mingjie Jiang, Xinyi Hu, Yuxin Liu, Bo Cao, Jiuxin Southeast Univ Nanjing Peoples R China

ISBN: (纸本)9789819794423;9789819794430

The development of Large Vision-Language Models (LVLMs) has been hindered by hallucinations. Existing methods often struggle to accurately infer relationships between objects and their attributes, and frequently overlook the challenges posed by semantic duality. In this work, we develop METER, a novel multimodal hallucination detection method that utilizes a mixture of experts through tool-supported reasoning and ensembling. Specifically, our model rethinks and infers based on the decomposed reasoning steps derived from the chain-of-thought prompts, which eliminates the need for additional manual templates and recognizes attributes of hallucination step by step. We also use topics discovered from image-text pairs to distinguish ambiguous text, mitigating semantic duality. Furthermore, we investigate the effects of incorporating external tools into hallucination detection, exploring the variations and efficacy of tool ensembling in mitigating hallucinations. Additionally, we successfully alleviate hallucinations by incorporating METER's explanation into the prompt. Extensive experiments demonstrate the effectiveness of our model. Our codes are available at https://***/lambdarw/METER.

关键词： Hallucination Detection LVLMs Tools Integration CoT

来源：评论

学校读者我要写书评

暂无评论

Web search result diversification by combining global and local document features

引用

APPLIED SOFT COMPUTING 2025年 169卷

作者： Zhu, Rui zhang, Shengxiang Liu, Bo Tian, Qiuyu Wu, Xiaoqing zhang, ruwen Cao, Jiuxin Qian, Linfeng Southeast Univ Sch Cyber Sci & Engn Nanjing Peoples R China Southeast Univ Sch Comp Sci & Engn Nanjing Peoples R China Jiangsu Marine Econ Monitoring & Evaluat Ctr Nanjing Peoples R China

Current search engines playa crucial role in information retrieval, shaping how we access and interact with the vast sources of knowledge available online, yet effectively handling ambiguous or broad user queries remains a significant challenge, as these engines typically rank web documents based primarily on relevance. While search result diversification is acknowledged as a key method for enhancing the alignment of results with a user's implicit intent, existing methods tend to focus on either a global or local perspective of document retrieval. These approaches commonly overlook the dynamic interactions between selected and candidate documents in global strategies and fail to capture the intricate internal correlations among all documents in local assessments. This oversight sets the stage for our proposed MultiViewDIV model, which aims to improve search result diversification through the integration of multi-view feature fusion. MultiViewDIV combines both local and global document diversity features with relevance features between documents and queries. For local diversity, it constructs a novel document relationship graph that incorporates temporal correlations to better understand the inter-document relationships. For global diversity, it introduces a method to assess the dissimilarity among documents and the extent to which they capture the implicit search intent of the query. The model fuses both global and local diversity features into a multilayer perceptron to compute document scores, which encapsulate diversity and relevance of each document, are then used to rank the documents in an optimal manner. Extensive experiments on two public datasets confirm the effectiveness and superiority of MultiViewDIV over existing approaches, achieving relative improvements of 1.07% in alpha-nDCG, 1.08% in ERR-IA, 0.89% in NRBP, and 1.62% in Pre-IA.

关键词： Document graph Global diversity Local diversity Search result diversification

来源：评论

学校读者我要写书评

暂无评论

Using the principles of Exploratory Practice to guide group work in an extensive reading class in China

引用

Language Teaching Research 2004年第3期8卷 331-345页

作者： zhang, ruwen Zhejiang University of Finance and Economics China

Exploratory Practice emphasizes integrating research into pedagogy, and attaches great importance to the quality of life in the classroom. It suggests that we work primarily to understand language classroom life, to bring people together, and to develop students’ language competency in a harmonious *** putting EP principles into my teaching practice, I explore ways of solving a problem I experienced in teaching English extensive reading to English majors in a Chinese University. © 2004, Sage Publications. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Building a Solid Bridge between Chinese and Western Cultures: Commentary on Wu Di's Comparative Studies in Poetry

INTERDISCIPLINARY STUDIES OF LITERATURE

引用

INTERDISCIPLINARY STUDIES OF LITERATURE 2024年第4期8卷 652-663页

作者： zhang, ruwen Zhejiang Univ Finance & Econ Sch Foreign Languages Hangzhou 310018 Peoples R China

Wu Di's comparative studies of Chinese and Western poetry focus on two interrelated considerations: the distinctive artistic appeal of poetic texts and the cultural exchanges between the East and the West. These exchanges are evident in the mutual appreciation of civilizations, which is reflected in the significant artistic discipline of poetry. In his monographs, such as Comparative Studies in European and American Poetry and Studies on English Metaphysical Poetry, Wu Di explores important issues in British poetry and examines the subtle relationship between Chinese and Western poetry, along with its complex influences, all showcased through his keen philosophical thinking and exquisite writing style. Furthermore, he offers an insightful account of the structural and artistic differences between Chinese and Western poetry, as well as their spiritual integration, particularly through the practice of poetry translation. Wu Di has made a commendable effort to build a solid bridge between Chinese and Western cultural exchanges by studying poetic texts from both traditions, which deserves recognition and praise.

关键词： Wu Di Chinese and Western poetry comparative studies cultural exchanges

来源：评论

学校读者我要写书评

暂无评论

Using online negative emotions to predict risk-coping behaviors in the relocation of Beijing municipal government

引用

SCIENTIFIC REPORTS 2024年第1期14卷 1-15页

作者： Xie, Qihui Wu, Hongyu zhang, ruwen China Univ Min & Technol Beijing Sch Law & Humanities Dept Publ Adm Beijing 100083 Peoples R China Beijing Forestry Univ Sch Art & Design Beijing 100083 Peoples R China

This article explores the use of online negative emotions to predict public risk-coping behaviors during urban relocation. Through a literature review, the paper proposes hypotheses that anticipate advanced prediction of public risk-coping behaviors based on online negative emotions. The study's empirical focus is on the relocation of the Beijing municipal government, using time series data for Granger causality analysis in EViews 10.0 software. Data on online negative emotions is sourced from Sina Weibo. After data cleaning, 1420 pieces of data related to the relocation policy of the Beijing Municipal Government within the period from June 9, 2015 to April 28, 2019 are retained. while risk-coping behaviors are measured through public information search behaviors and the incidence of violent crimes, the data coverage is also from June 9, 2015 to April 28, 2019. The results indicated that: (1) Online negative emotions regarding the relocation policy predict public risk-coping behaviors in advance. (2) Negative comments are more effective predictors than negative feelings;(3) Negative emotions about relocation policy formulation predict risk-coping behaviors better than those related to policy effectiveness and implementation;(4) Negative emotions from individuals better predict public risk-coping behaviors than those from institutions;(5) Negative emotions from key stakeholders better predict public risk-coping behaviors than those from non-key or marginal stakeholders. It is recommended that relevant departments establish a real-time monitoring system to track negative public opinions and emotions expressed online, adopt a stakeholder-centric approach to facilitate communication, and promote transparency and educational campaigns to address the challenges of urban relocation. In future studies, methods such as expanding the sample size and adding indicators will be used to address the limitations of potential bias in sample data.

关键词： Negative emotions online Information search behavior Violent behavior Granger causality analysis Relocation of Beijing municipal government

来源：评论

学校读者我要写书评

暂无评论

Wetland types shape microbial function related to soil phosphorus cycling based on metagenomics analysis

引用

Journal of Environmental Sciences 2025年

作者： Meiqing Jia Zhiwei Gao Xuefeng Hu Mengqiong Feng Fanhui Liu Jing Huang ruwen zhang Jing Li Guogang zhang Yidong Wang Tianjin Key Laboratory of Water Resources and Environment Tianjin Normal University Tianjin 300387 China Department of Geography Tianjin Normal University Tianjin 300387 China College of Life Sciences Tianjin Normal University Tianjin 300387 China Tianjin Key Laboratory of Animal and Plant Resistance Tianjin Normal University Tianjin 300387 China Journal Publishing Center Tianjin Normal University Tianjin 300387 China

Limitations regarding phosphorus (P) are widespread in ecosystems. Understanding the impacts of the wetland types on microbially mediated soil P availability and cycling is essential for the effective management of wetlands. In this study, the Beidagang wetland (B), Baodi paddy field (R), and Dahuangpu wetland (W) in Tianjin, China were chosen as representatives of the coastal, constructed, and swampy wetland, respectively. Sequential P extraction and metagenomics approaches were adopted to explore the soil P fraction and microbially regulated P cycle. Proteobacteria were the predominant microbes-related soil P cycle. IMPA, gph, rsbU_P, ugpQ , and glpK genes were dominant in organic P (Po) mineralization, while gcd, ppa , and ppx genes were dominant in inorganic P (Pi) solubilization. The salinity, NO 3 - -N concentration, the ratio of total carbon to total nitrogen (TC/TN), total carbon (TC), and the ratio of soil organic carbon to total P (SOC/TP) were the co-drivers of microbially mediated P cycle processes. Microbial network complexity-relate P cycle was the lowest in the coastal wetland. Salinity and NO 3 - -N exhibited a significant negative relation to the abundance of most genes-relate Pi solubilization and a remarkable positive correlation with the abundance of many genes-relate Po mineralization. These findings demonstrated that Po mineralization tended to occur in habitats with high salinity and nutrient imbalances, whereas the dissolution of Pi was prone to occur in low-salinity environments with relatively balanced soil nutrients. This study improves understanding of how salinity and soil nutrients jointly shape microbial-regulated soil P cycle in different types of wetlands.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Weakly-supervised structural component segmentation via scribble annotations

引用

COMPUTER-AIDED CIVIL AND INFRASTRUCTURE ENGINEERING 2025年第5期40卷 561-578页

作者： zhang, Chenyu Li, Ke Yin, Zhaozheng Qin, ruwen SUNY Stony Brook Dept Civil Engn Stony Brook NY 11794 USA SUNY Stony Brook Dept Comp Sci Stony Brook NY USA SUNY Stony Brook Dept Biomed Informat Stony Brook NY USA SUNY Stony Brook AI Inst Stony Brook NY USA

Segmentation of structural components in infrastructure inspection images is crucial for automated and accurate condition assessment. While deep neural networks hold great potential for this task, existing methods typically require fully annotated ground truth masks, which are time-consuming and labor-intensive to create. This paper introduces Scribble-supervised Structural Component Segmentation Network (ScribCompNet), the first weakly-supervised method requiring only scribble annotations for multiclass structural component segmentation. ScribCompNet features a dual-branch architecture with higher-resolution refinement to enhance fine detail detection. It extends supervision from labeled to unlabeled pixels through a combined objective function, incorporating scribble annotation, dynamic pseudo label, semantic context enhancement, and scale-adaptive harmony losses. Experimental results show that ScribCompNet outperforms other scribble-supervised methods and most fully-supervised counterparts, achieving 90.19% mean intersection over union (mIoU) with an 80% reduction in labeling time. Further evaluations confirm the effectiveness of the novel designs and robust performance, even with lower-quality scribble annotations.

关键词： Deep neural networks

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：