Scholarlink

<p>Freshwater systems and urban water distribution networks are increasingly exposed to pollution, infrastructure failures, and overuse. Timely geospatial information is needed to detect emerging problems and support rapid intervention, yet conventional monitoring is often costly, slow, and spatially incomplete. This paper presents an ontology-guided spatial crowdsourcing framework that transforms volunteered reports into semantically consistent spatio-temporal event candidates for water resources (WRs) and water distribution systems (WDSs). A lightweight, multi-level domain ontology is distilled into volunteer-facing conceptual models for structured reporting, and a lookup table links user selections back to ontology classes to support semantic alignment at scale. Event discovery is performed via ontology-constrained spatio-temporal clustering (DBSCAN) using a composite similarity that integrates textual, spatial, temporal, and ontological proximity. The framework was implemented in a web-based system and evaluated in a three-month campaign involving 572 volunteers and 1,019 ontology-structured reports, with performance assessed against available field verification records. The pipeline achieved Precision = 0.86, Recall = 0.91, and F1 = 0.89. Spatial and temporal alignment errors were limited (median spatial error ≈ 12&#xa0;m; median temporal error ≈ 1&#xa0;h), and discovery latency was typically short. Compared with post-hoc text parsing or flat labeling, the “semantics-at-source” design reduces terminology fragmentation and yields standardized event records that can be reviewed and integrated with operational GIS layers. The results indicate that ontology-guided spatial crowdsourcing can support coherent event discovery and triage for WR/WDS management and can be adapted to other risk and infrastructure domains.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An ontology-guided spatial crowdsourcing framework for spatio-temporal event discovery in water resources management

  • 期刊论文

  • 发表日期：

    2026年4月10日

查看全文
  • Hossein Bahadorizadeh,
  • Mohammadreza Malek,
  • Liqiu Meng

摘要

Freshwater systems and urban water distribution networks are increasingly exposed to pollution, infrastructure failures, and overuse. Timely geospatial information is needed to detect emerging problems and support rapid intervention, yet conventional monitoring is often costly, slow, and spatially incomplete. This paper presents an ontology-guided spatial crowdsourcing framework that transforms volunteered reports into semantically consistent spatio-temporal event candidates for water resources (WRs) and water distribution systems (WDSs). A lightweight, multi-level domain ontology is distilled into volunteer-facing conceptual models for structured reporting, and a lookup table links user selections back to ontology classes to support semantic alignment at scale. Event discovery is performed via ontology-constrained spatio-temporal clustering (DBSCAN) using a composite similarity that integrates textual, spatial, temporal, and ontological proximity. The framework was implemented in a web-based system and evaluated in a three-month campaign involving 572 volunteers and 1,019 ontology-structured reports, with performance assessed against available field verification records. The pipeline achieved Precision = 0.86, Recall = 0.91, and F1 = 0.89. Spatial and temporal alignment errors were limited (median spatial error ≈ 12 m; median temporal error ≈ 1 h), and discovery latency was typically short. Compared with post-hoc text parsing or flat labeling, the “semantics-at-source” design reduces terminology fragmentation and yields standardized event records that can be reviewed and integrated with operational GIS layers. The results indicate that ontology-guided spatial crowdsourcing can support coherent event discovery and triage for WR/WDS management and can be adapted to other risk and infrastructure domains.