本論文提出雲端農藥數據處理系統，依據農藥數據處理演算法與網路爬蟲配合虛擬化技術，對應每個儀器產出的樣品數據進行數據處理，將每個樣品的農藥分為合格以及不合格兩種結果。
農藥數據處理演算法，裡頭包含”品質管理”演算法以及”數據篩選”演算法，每批要數據處理的樣品都會有一套品管資訊，透過品質管理演算法將這些樣品的品質管理資料先進行資料整合及運算，再與系統給予的品質管理標準進行篩選與比對，便可知道該批樣品的品質管理是否合格，樣品的品質管理處理完後，再把每個樣品透過數據篩選演算法來過濾出陽性驗出的農藥；網路爬蟲的工作主要是透過網路爬取政府公告的推薦用藥登錄資料及農藥殘留容許量標準，把過濾出來陽性驗出的農藥與網路爬蟲爬取的標準進行比對，便可知道該樣品是否有符合政府所提供的農藥殘留標準。
虛擬化技術其中一項特色就是可以隨著服務的需求，去制定資源耗費的使用量，並且還能夠擴充資源來滿足系統需求，能讓實體主機的資源透過虛擬化技術來妥善運用，本系統使用的是Docker以及Cloudstack所建置的容器Container以及虛擬機器Virtual Machine，Docker引擎上所建立的容器，主要負責的部分是系統的演算法及網路爬蟲，Cloudstack則是建立資訊與數據的資料庫，將上傳的數據及系統處理完的數據儲存至具有資料庫之虛擬機器裡頭。
而Docker沒有額外運行作業系統Guest Operating System，故所建立的容器映像檔Image的檔案容量，由於減少了作業系統所需的儲存空間，因此即便安裝許多軟體或應用程式，容器映像檔Image也僅需約數百萬位元組(MB)，在映像檔的管理及服務的啟動、關閉流程都比虛擬機器所建立的映像檔來的快速許多，也因為Docker的架構，無論在任何作業系統下，只要建置Docker Engine，就可以利用容器映像檔來啟用容器，換言之，行動裝置（如：筆記型電腦）在不同作業系統情況下，利用Docker Engine可以直接移植容器Container，使得服務可以成為可攜式的服務。
雲端運算可以根據使用者的需求透過網際網路來啟用服務，也因為服務都架設在伺服器上，所以雲端運算利於管理、維護、開發服務，搭配虛擬化技術能更加有效的去運用資源池。
因為系統以及儲存的資料都在虛擬機器裡，因此透過雲端管理與運算資源池，資料庫儲存空間可以彈性調整，所以系統維護人員能夠快速又方便的維護此系統，也因為系統建置在雲端平台上，檢驗員可透過各種的遠端操作系統連線至Server進行操作，且本研究透過網際網路提供遠端存取，多樣性服務需要作業系統管理，並用農藥數據處理演算法與網路爬蟲來建構容器的資料內容，因此於本研究中，利用雲端虛擬機器作為伺服端。
基於本研究所提出之雲端農藥數據處理系統，對於伺服器的管理可以有效的利用資源池，且透過容器Container，可以達到輕量化、移植性高的成效。而且整體系統可以縮減產出農藥檢測報告的時間，也可以降低人為操作的失誤。

This paper proposed a Cloud Agriculture Safety Data Process System. Using a pesticide data processing algorithm based on ther web-crawler and virtualization technology, sample data collected by specific instruments could be categorized into 2 groups, qualified and unqualified.
This pesticide data processing algorithm includes the quality management algorithm and the data filtering algorithm. Every data set includes a number of quality insurance information. The quality management algorithm will organize the information before processing, filtering, and cross-referencing. Then, considering the quality insurance data provided by the system, the samples will be determined whether is qualified for the market or not. This pesticide data processing algorithm will also filter out each pesticide found in each data.
The main purpose of Web-Crawler included in this system is to gather national regulation of each pesticide. By comparing the pesticide data which is found by the pesticide data processing algorithm with these regulations, the proposed agriculture safety data process system can qualify and verify the safety of the agriculture.
Virtualization allows the user to not only customize system resources according to the service, but also easily add resources to match system requirements. In this research, the proposed system utilize Containers formed by Docker and Virtual Machines managed by Cloudstack. Containers formed using Docker Engine mainly handle the algorithms and web-crawler; with Cloudstack focusing on constructing the database in a virtual machine using data processed by the system.
Due to Docker not needing an additional Guest Operating System, the file size is significantly smaller than the traditional virtual machines. Even with plenty of software and application installed, the image file of one virtual machine usually just consumes hundreds MegaBytes storage. Furthermore, it is much more responsive to start, drop, and mange docker-container images. Based on Docker’s infrastructure, Docker images could be implemented on any Operating System with Docker Engine installed.
Since the data in this system is stored in a virtual machine, storage space is easily adjustable via resource pool. By implementing the system on a cloud platform, users can access the proposed system remotely by using any. In order to achieve this function, along with needing an operating system for running multiple services, and to construct the data for container via the pesticide data processing algorithm and web-crawler. A cloud-based virtual machine was chosen as the server for the proposed system.
In the Pesticide data processing system proposed in this paper, cloud environment is affectively used due to the resources. It is very light and highly portable due to the use of Docker Containers. Finally, it significantly cut down the time needed to obtain pesticide reports and the chance of a human mistake.

中文摘要 …………………………………………………………... i
英文摘要 …………………………………………………………... iii
誌謝 …………………………………………………………... v
目錄 …………………………………………………………... vi
表目錄 …………………………………………………………... vii
圖目錄 …………………………………………………………... viii
符號說明 …………………………………………………………... ix
第一章 緒論…………………………………………………....... 1
第二章 文獻探討………………………………………………... 3
2.1 農藥檢測………………………………………………... 3
2.2 網路爬蟲………………………………………………... 4
2.3 雲端運算………………………………………………... 5
2.4 虛擬化…………………………………………………... 5
第三章 研究內容與方法………………………………………... 7
3.1 農藥數據處理演算法...……………………………….... 9
3.2 網路爬蟲...…………………………………………….... 11
第四章 實驗部份………………………………………………... 13
第五章 結論……………………………………………………... 17
參考文獻 …………………………………………………………... 18
英文論文大綱 …………………………………………………………... 19
簡歷 …………………………………………………………... 24

[1]登記用藥查詢. https://pesticide.baphiq.gov.tw/web/Insecticides_MenuItem5_3.aspx
[2]法規內容. https://pesticide.baphiq.gov.tw/web/LawsDetailView.aspx?law_code=1&sn=103
[3]食品藥物消費者知識服務網-殘留農藥安全容許量標準. https://consumer.fda.gov.tw/Law/PesticideList.aspx?nodeID=520
[4]簡銘伸，雲端作業系統，金琅學術出版社, 2016
[5]Chandni Saini, Vinay Arora, “Information retrieval in web crawling: A survey,” Advances in Computing, Communications and Informatics (ICACCI), 2016 International Conference on.
[6]質譜法. https://zh.wikipedia.org/wiki/%E8%B4%A8%E8%B0%B1%E6%B3%95
[7]食品中殘留農藥檢驗方法-多重殘留分析方法. www.fda.gov.tw/tc/includes/GetFile.ashx?mID=19&id=64555
[8]Ming-Shen Jian, Hao-Yi Xu, Jenn-Feng Sheen, Yi-Ling Ye, “Cloud based agriculture safety inspection with multiple standard sources,”Advanced Communication Technology (ICACT), 2018 20th International Conference on.
[9]農業藥物毒物試驗所. https://www.tactri.gov.tw/