以語意網為方式之企業氣候預警通知系統

現今資訊科技發展蓬勃，變化最多的莫過於網際網路中資料的應用層面愈來愈多元，而資料量隨著時間的演進變得非常龐大，帶來的問題即是資料多而雜亂。目前在資料檢索上還是靠著一般的關鍵字檢索技術居多，以氣候、災害的領域來說，在茫茫的網路中會檢索到很多不同的資料，導致一般使用者面臨過多資料而無法快速、有效的尋找到真正所需的知識；而對於研究氣候災害的人員要查詢來說也是具有相同問題。在國內的氣候資訊系統中所提供之資訊，大多缺乏提供專門對企業、廠商所設計的預警資訊。故本研究為了能夠讓使用者能方便快速的查詢到想要的氣候、災害知識和解決企業在氣候災害上損失的問題，研製了一個具語意檢索及預警通知之企業氣候調適資訊系統，在系統中利用網路爬蟲技術擷取了政府之氣候、災害相關資訊網站上的公開資訊以及政府推廣之氣候開放資料（open data），經過整合處理及友善的資訊呈現平台，能夠讓使用者快速找尋到自己想要的資訊；系統建置的語意檢索功能搭配著建置的氣候、災害本體論資料庫和語意推論引擎，可分析使用者輸入的相關查詢句詞，提供以詞彙句子的問答式檢索來查詢氣候、災害、氣候調適等知識，藉此來達到人性化的問答檢索功能，提升獲取氣候資訊的準確性與效率。而系統中建置之判讀技術會使用氣候預警規則來分析即時的氣候資訊，若達到危機警報條件，則馬上送出預警通知至企業廠商，讓企業達到提早防災之準備，以減少氣候災害所帶來的巨大損失。

關鍵字

語意網；本體論；語意檢索；氣候調適；警訊通

並列摘要

At present, information science and technology are booming, volume of data from Internet have emerged very huge. With the evolution time, we face the problem is data processing became more complex. Currently, the information retrieval technology is still relying on a general keyword search technology mostly. In the area of climate and disasters, querying from network will acquire a lot of different information, it makes the general users get excessive data but not needing. Many climate information systems are lack of warning design for enterprises. For the purpose of providing weather information easily and quickly for users, and to solve problems of business losses about climate disaster. We develop a corporate climate adaptation information system with the function of semantic retrieval and warning notification. The system uses a network crawler to capture real-time weather information from major climate information websites, and collecting the open data providing by the government. We design a friendly information presentation platform that allows users to find information what they want quickly. The semantic query function of system with climate disasters ontology repository and semantic inference engine analyzes related queries sentences those enter by users. All sentences are typed in vocabulary quiz-form and query about weather, disasters, climate adaptation etc. knowledge. By continuous Q&A operation between users and UI response, enhancing the accurate of access on climate information. The interpretation techniques is the use of climate warning rules to reason climate information immediately. If climate situation reaches the alert notification conditions, warning notices will be sent to the enterprise immediately, and reaching the effectiveness of early disaster and reducing business losses about climate disaster.