GeoKG-HSA: A framework for habitat suitability assessment with geospatial knowledge graphs
文献类型：期刊论文
| 作者 | Xiao, Xin1; Wang, Peng1; Ge, Yong1,3; Luo, Jin1; Chen, Hao1; He, Yufeng1; Zhang, Die1; Li, Yankuo4; Fang, Chaoyang1; Lin, Hui1,2 |
| 刊名 | INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
 |
| 出版日期 | 2025-11-01 |
| 卷号 | 144页码:104921 |
| 关键词 | Habitat suitability assessment Geospatial knowledge graph Environmental modelling Knowledge-driven Machine learning |
| ISSN号 | 1569-8432 |
| DOI | 10.1016/j.jag.2025.104921 |
| 产权排序 | 3 |
| 文献子类 | Article |
| 英文摘要 | Habitat suitability, which reflects how well environmental conditions support species survival, growth, and reproduction, is highly sensitive to climate change and human activities and underpins effective conservation policy-making. Traditional machine learning models for habitat suitability assessment often struggle to capture complex semantic relationships among environmental factors and suffer from limited spatial generalization. To overcome these limitations, we propose GeoKG-HSA, a novel framework leveraging geospatial knowledge graphs to enhance habitat suitability assessment. By adopting an ontology-based approach, GeoKG-HSA explicitly models semantic relationships among diverse environmental variables, enabling structured integration of multisource heterogeneous habitat data. The framework further integrates symbolic domain knowledge with datadriven machine learning, establishing a hybrid paradigm that improves both interpretability and predictive accuracy. A case study on Tundra Swan habitat in a wetland ecosystem shows that, on average across four models, GeoKG-HSA improves precision by 0.147, recall by 0.281, and F1 score by 0.255. Area Under the Curve (AUC) scores were also improved by 0.04 for the logistic regression model, 0.20 for the support vector machine model, 0.06 for the decision tree model, and 0.13 for the multi-layer perceptron model. Compared with the MaxEnt, GeoKG-HSA exhibits superior spatial generalization in regions with missing or sparse samples. These results demonstrate that GeoKG-HSA effectively integrates multi-source environmental data, enhances model generalizability, and significantly outperforms traditional Tabular-HSA methods. This framework offers a promising tool for advancing biodiversity conservation and ecosystem management under changing environmental conditions. |
| URL标识 | 查看原文 |
| WOS关键词 | ONTOLOGY ; ENRICHMENT |
| WOS研究方向 | Remote Sensing |
| 语种 | 英语 |
| WOS记录号 | WOS:001619278100004 |
| 出版者 | ELSEVIER |
| 源URL | [http://ir.igsnrr.ac.cn/handle/311030/217674]  |
| 专题 | 资源与环境信息系统国家重点实验室_外文论文 |
| 通讯作者 | Ge, Yong; Lin, Hui |
| 作者单位 | 1.Jiangxi Normal Univ, Sch Geog & Environm, Key Lab Poyang Lake Wetland & Watershed Res, Minist Educ, Nanchang 330022, Peoples R China; 2.Jiangxi Normal Univ, Nanchang Base Int Ctr Space Technol Nat & Cultural, Nanchang 330022, Peoples R China; 3.Inst Geog Sci & Nat Resources Res, Chinese Acad Sci, Beijing 100101, Peoples R China; 4.Jiangxi Normal Univ, Coll Life Sci, Nanchang 330022, Peoples R China |
| 推荐引用方式 GB/T 7714 | Xiao, Xin,Wang, Peng,Ge, Yong,et al. GeoKG-HSA: A framework for habitat suitability assessment with geospatial knowledge graphs[J]. INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION,2025,144:104921. |
| APA | Xiao, Xin.,Wang, Peng.,Ge, Yong.,Luo, Jin.,Chen, Hao.,...&Lin, Hui.(2025).GeoKG-HSA: A framework for habitat suitability assessment with geospatial knowledge graphs.INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION,144,104921. |
| MLA | Xiao, Xin,et al."GeoKG-HSA: A framework for habitat suitability assessment with geospatial knowledge graphs".INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION 144(2025):104921. |
入库方式： OAI收割
来源：地理科学与资源研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。