在發展中國家城市中警察工作是一項艱鉅的任務,而社會中普遍人民仍對警寄 與高度期望即使社會資源如此匱乏,這些壓力加深了社會中許多的零星犯罪, 導致警察們經歷一段密集的工作時期,這也讓他們沒有多餘的時間來建立精準 毫無錯誤的報告。而報告數據中的錯誤以及資料不一致也讓在對這些數據進行 地理編碼時更加困難,這也進一步造成大多數事件報告仍然無法進行地理編 碼。然而我們可以透過模糊集合理論找出每個事件報告的關係來緩解這個問 題。我們常利用地理編碼數據與無法做編碼的數據之間的關係來生成相近的事 件位置。在本篇論文中,針對不可進行地理數據編碼的資料藉由找出地形特 徵,時間特徵和警察局的位置資訊,來近似估出犯罪事件的所在位置,並可以 讓以群集方式產生的地理編碼事件更為豐富。
Police work can be a difficult task in the urban cities of developing nations, high expectations combined with a lack of resources are common occurrences. These stresses are further compounded by the sporadic nature of crime, causing officers to experience periods of intense work activity. As a result officers spend a very small amount of their available time to ensure flawless report creation. Errors in report data coupled with inconsistent representations make geocoding this data very difficult. These difficulties causes the majority of incident reports to remain ungeocodable, and by extension unusable for clustering. However, this problem is mitigated through the application of fuzzy set theory, relationships between incident reports can be formed. Relationships between geocodable data and ungeocodable data are used to generate an approximation of the ungeocodable incident’s location. In this thesis the relationships found in topographic features, temporal features and the modeling of police officer information are used to generate approximate location information for ungeocodable crime incidents. Which can then be used to enrich geocoded incidents in crime cluster generation.