地理網路 (GeoWeb) 指的是任何包含地理資訊的網際網路(World-Wide Web, WWW)資源，其中資源(resource)可以是個別資料(data)或網路服務(web service)。透過網際網路，使用者可便利地取得地理網路資源並在許多領域應用，然而因為資源由不同的個別用戶或機構，基於各自的資料模型或編碼，地理網路有著嚴重的異質性問題(heterogeneity issues)：包括地理網路缺乏一致的模型，以及無法顯示資源間隱含的語意(semantic)關係，所以用戶需要了解不同的資料模型和各個資料提供者的表達架構，以探索並識別他們感興趣的地理網路資源；此外，地理網路資源之間的隱含的語意關係，導致簡單的關鍵字匹配不足以讓用戶尋找到其需的地理網路資源。因此，本研究旨在透過完整且易於理解的知識本體（ontology），同時根據不同標準自動描述各種類型的地理網路資源來解決上述問題。而除了考慮了不同的資料模型，本研究亦考慮了地理網路資源間關於主題的隱含語意關係：我們透過WordNet計算語意相似度(semantic similarity)，解析詮釋資料(metadata)，並據此自動化統一以資源描述框架 (Resource Description Framework, RDF)儲存地理網路資源，最後透過SPARQL協定與RDF查詢語言(SPARQL Protocol and RDF Query Language, SPARQL Query)搜索資源，即可透過此研究所提出的知識本體，發現地理網路資源間的主題隱含之語意關係。從結果我們可以看到，知識本體配合語意相似度地計算，可以自動地整合地理網路資源的異質性問題，促進地理網路資源的發現及應用。;The Geospatial Web (GeoWeb) represents the collection of World-Wide Web resources containing geographic information, where resources could be individual data or web services. With the help of the Internet, GeoWeb resources are easily accessible and have been applied in many fields. However, GeoWeb faces serious heterogeneity issues as resources are provided by various individuals or organizations and are usually based on different data models or encodings. Among the heterogeneity issues, the semantic heterogeneity issue includes lack of a coherent conceptual model for GeoWeb resources and hidden relationships between resources. Hence, users need to understand different data models and individual ontologies to identify and discover the GeoWeb resources of their interest. Furthermore, implicit relationships between GeoWeb resources cause problems on simple keyword matching as users may not search with exact keywords. Therefore, this study aims at addressing the aforementioned issues by describing the GeoWeb resources based on a complete and accessible ontology. Additionally, automatic matching various types of geospatial resources according to each standard. This research considers not only dissimilar data models but also the implicit semantic relationships between their thematic metadata. With WordNet, we parsed metadata of resources and calculated the semantic similarity between proposed ontology and metadata to discover their implicit relationships. With resources represented in the Resource Description Framework (RDF) format, SPARQL queries can be issued to search for resources based on their semantic relationships. The result shows that the proposed solution can integrate heterogenous GeoWeb resources and facilitate the GeoWeb resource discovery.
