環保專案成果報告查詢系統

專案計畫編號
經費年度	111
計畫經費	1,970 千元
預算科目	數位建設-建構開放政府及智慧城鄉服務
專案開始日期	2022/04/26
專案結束日期	2022/11/30
專案主持人	柯上茗
主辦單位	環境督察總隊
承辦人	蕭一川
執行單位	環資國際有限公司
專案分類	區域環境管理
中文關鍵字	督察，環保業務，視覺化、事件特性標籤、金融資料分析
英文關鍵字	Inspection, environmental protection operation, visualization, event characteristic label, financial data analysis
協同主持人	徐淨慧
共同主持人
計畫聯絡人	尤映文
計畫聯絡電話	02-23611999#844
計畫聯絡信箱	vegetable112@eri.com.tw

中文摘要	本年度建置督察視覺化資訊平臺執行主要成果如下： (一)研析建構模組智能化視覺多點中心作法 1.視覺化平臺的開發方式已使用響應式網頁 (RWD)，以提供使用者在各式裝置上可正常使用視覺化平臺。 2.研析許可及申報資料，並分析出可運用的資料，呈現在視覺化平臺上。 3.依據督察實際執行情形，將特別關注議題如：疑似繞流、暗管排放熱區、稽查人員派案追蹤、行業別聚落分布、應加強管理之工業區及廢棄物清除業者收受分布，併同事業機構相關資料及關切產業聚落熱點分析呈現於視覺化平臺，作為「七大主題視覺化儀表板」。 4.在視覺化平臺上呈現相關主題時，採用地理空間地圖做為主要視覺畫面，並以事業點位及基本資料做為展示方式。 5.於系統查詢介面設定行業別、廢棄物代碼、原物料代碼、工業區、流域別做為搜尋條件，並以事業點位呈現搜尋結果。 6.以歷年督察人員經手的稽查案件數量及投入總人次為基礎，以縣市及污染別為查詢條件．以利掌控各縣市督察狀況。 7.利用AI技術將半結構化的欄位資料進行分詞處理後，以自動化查找方式擷取及搜尋金融相關資料；並以七大主題視覺化儀表板為軸心，提出AI及IoT 之應用規劃方法。 (二) 研析特定督察案件樣態連結方式：針對特定污染類型督察案件（例如：以事業、量、關注項目、時或地等類型）資料的連結作法，其中連結內容需包含以下資料內容 1.盤點事業歷年查處資料，擷取重點精華資訊，標定為事件特性標籤，並將標定的61組標籤以機器學習的方式建立模型，共訓練得出34組模型。 2.將34組模型利用社會網絡分析圖(SNA)展示同一個事件特性標籤關聯的事業機構，或是單一事業機構所標記各個污染別的事件特性標籤。 (三) 研析利用「金融資料」輔助督察之可行性 1.利用自動爬蟲方式擷取開放式金融資料，建立上市櫃公司之歷史金融數據資料庫，以金融資料比對水系統申報資料，共建立1,404間事業資料表。 2. 利用4種金融與6種水系統申報數據，建立24組比對模式，採用迴歸分析方式找出24種比對模式中金融與水污染申報數據相關係數較高者，列為具參考性的因子。 (四) 機關交辦其他相關事項 1.配合完成機關交辦事項及定期召開工作進度會議、參加計畫執行情形會報、辦理交付文件撰寫。 2.已協助辦理於記者發表會上展示系統畫面，持續配合機關需求。
中文關鍵字	督察，環保業務，視覺化、事件特性標籤、金融資料分析

中文摘要

本年度建置督察視覺化資訊平臺執行主要成果如下： (一)研析建構模組智能化視覺多點中心作法 1.視覺化平臺的開發方式已使用響應式網頁 (RWD)，以提供使用者在各式裝置上可正常使用視覺化平臺。 2.研析許可及申報資料，並分析出可運用的資料，呈現在視覺化平臺上。 3.依據督察實際執行情形，將特別關注議題如：疑似繞流、暗管排放熱區、稽查人員派案追蹤、行業別聚落分布、應加強管理之工業區及廢棄物清除業者收受分布，併同事業機構相關資料及關切產業聚落熱點分析呈現於視覺化平臺，作為「七大主題視覺化儀表板」。 4.在視覺化平臺上呈現相關主題時，採用地理空間地圖做為主要視覺畫面，並以事業點位及基本資料做為展示方式。 5.於系統查詢介面設定行業別、廢棄物代碼、原物料代碼、工業區、流域別做為搜尋條件，並以事業點位呈現搜尋結果。 6.以歷年督察人員經手的稽查案件數量及投入總人次為基礎，以縣市及污染別為查詢條件．以利掌控各縣市督察狀況。 7.利用AI技術將半結構化的欄位資料進行分詞處理後，以自動化查找方式擷取及搜尋金融相關資料；並以七大主題視覺化儀表板為軸心，提出AI及IoT 之應用規劃方法。 (二) 研析特定督察案件樣態連結方式：針對特定污染類型督察案件（例如：以事業、量、關注項目、時或地等類型）資料的連結作法，其中連結內容需包含以下資料內容 1.盤點事業歷年查處資料，擷取重點精華資訊，標定為事件特性標籤，並將標定的61組標籤以機器學習的方式建立模型，共訓練得出34組模型。 2.將34組模型利用社會網絡分析圖(SNA)展示同一個事件特性標籤關聯的事業機構，或是單一事業機構所標記各個污染別的事件特性標籤。 (三) 研析利用「金融資料」輔助督察之可行性 1.利用自動爬蟲方式擷取開放式金融資料，建立上市櫃公司之歷史金融數據資料庫，以金融資料比對水系統申報資料，共建立1,404間事業資料表。 2. 利用4種金融與6種水系統申報數據，建立24組比對模式，採用迴歸分析方式找出24種比對模式中金融與水污染申報數據相關係數較高者，列為具參考性的因子。 (四) 機關交辦其他相關事項 1.配合完成機關交辦事項及定期召開工作進度會議、參加計畫執行情形會報、辦理交付文件撰寫。 2.已協助辦理於記者發表會上展示系統畫面，持續配合機關需求。

中文關鍵字

督察，環保業務，視覺化、事件特性標籤、金融資料分析

類型	檔名	檔案大小	說明	最新版公開日期
期末報告	111年督察資訊數位視覺化研析計畫_成果報告(定稿)部分公開版-刪除個資.pdf	2MB		2022/12/30 上午 12:00:00

摘要	In 2022, the results of establishing the Visualization Information Platform for inspection data are as follows: (A) Study and analyze the multiple primary methods for establishing modular intelligent visualization 1. The development method of the visualization platform used responsive web design (RWD) to provide users with appropriate visualization platforms on various devices. 2. This project studied and analyzed the permission and report data to select those to be used and presented on the visualization platform. 3. Based on actual inspections, this project presented special issues of concern, including suspected bypasses, hot spots of hidden pipe discharge, case tracking by inspectors, distribution of industry-specific clusters, industrial areas that should be further managed, and distribution of waste disposal enterprises, as well as relevant information from industry organizations and hot spots of relevant industrial clusters on the visualization platform as the Seven Thematic Visualization Dashboard. 4. When presenting the related themes on the visualization platform, the geospatial map was used as the main visualization screen, and the business locations and basic information were used as the display format. 5. Set industry, waste code, raw material code, industrial area, and watershed as search criteria in the system query interface, and the search results were presented by business locations. 6. Based on the number of inspection cases handled by inspectors and the total number of people who contributed to the system in the past years, the query criteria were determined based on counties, cities, and pollution categories to control the inspection status of each county and city. 7. AI technology was used to retrieve and search financial-related information by automating the word processing of semi-structured column data. The Seven Thematic Visualization Dashboard was used as the foundation to propose AI and IoT application planning methods. (B) Analyze the linkage method for specific inspection cases. This project analyzed the linkage method for specific types of pollution inspection cases (e.g. by business, volume, item of concern, time, or place). The linkage content should include the following data content: 1. This project inventoried the data of inspection and punishment in the past years, extracted the essential information and label them as event characteristics, and established models with 61 sets of labels by machine learning. A total of 34 models were obtained. 2. The 34 models were presented in a social network analysis (SNA) diagram to demonstrate the event characteristic labels associated with the same enterprise, or the event characteristic labels of each contamination marked by a single enterprise. (C) Analyze the feasibility of using financial data to assist inspection 1. This project used automatic web crawlers to retrieve open financial data to establish a database of historical financial data of listed companies. These data were compared with water system reporting data to create a total of 1,404 enterprise data tables. 2. A total of 24 sets of comparison patterns were established using four types of financial data and six types of water system reporting data. The relatively high correlation coefficients between financial data and water pollution reporting data among the 24 comparison patterns were identified by regression analysis and listed as the factors of interest. (D) Related tasks assigned by the authorities 1. This project cooperated with the authorities to complete the assigned tasks and organize regular progress meetings, participated in project implementation meetings, and completed designated documents. 2. This project assisted in the presentation of system screens at press conferences and continued to meet the needs of the authorities.
英文關鍵字	Inspection, environmental protection operation, visualization, event characteristic label, financial data analysis

摘要

In 2022, the results of establishing the Visualization Information Platform for inspection data are as follows: (A) Study and analyze the multiple primary methods for establishing modular intelligent visualization 1. The development method of the visualization platform used responsive web design (RWD) to provide users with appropriate visualization platforms on various devices. 2. This project studied and analyzed the permission and report data to select those to be used and presented on the visualization platform. 3. Based on actual inspections, this project presented special issues of concern, including suspected bypasses, hot spots of hidden pipe discharge, case tracking by inspectors, distribution of industry-specific clusters, industrial areas that should be further managed, and distribution of waste disposal enterprises, as well as relevant information from industry organizations and hot spots of relevant industrial clusters on the visualization platform as the Seven Thematic Visualization Dashboard. 4. When presenting the related themes on the visualization platform, the geospatial map was used as the main visualization screen, and the business locations and basic information were used as the display format. 5. Set industry, waste code, raw material code, industrial area, and watershed as search criteria in the system query interface, and the search results were presented by business locations. 6. Based on the number of inspection cases handled by inspectors and the total number of people who contributed to the system in the past years, the query criteria were determined based on counties, cities, and pollution categories to control the inspection status of each county and city. 7. AI technology was used to retrieve and search financial-related information by automating the word processing of semi-structured column data. The Seven Thematic Visualization Dashboard was used as the foundation to propose AI and IoT application planning methods. (B) Analyze the linkage method for specific inspection cases. This project analyzed the linkage method for specific types of pollution inspection cases (e.g. by business, volume, item of concern, time, or place). The linkage content should include the following data content: 1. This project inventoried the data of inspection and punishment in the past years, extracted the essential information and label them as event characteristics, and established models with 61 sets of labels by machine learning. A total of 34 models were obtained. 2. The 34 models were presented in a social network analysis (SNA) diagram to demonstrate the event characteristic labels associated with the same enterprise, or the event characteristic labels of each contamination marked by a single enterprise. (C) Analyze the feasibility of using financial data to assist inspection 1. This project used automatic web crawlers to retrieve open financial data to establish a database of historical financial data of listed companies. These data were compared with water system reporting data to create a total of 1,404 enterprise data tables. 2. A total of 24 sets of comparison patterns were established using four types of financial data and six types of water system reporting data. The relatively high correlation coefficients between financial data and water pollution reporting data among the 24 comparison patterns were identified by regression analysis and listed as the factors of interest. (D) Related tasks assigned by the authorities 1. This project cooperated with the authorities to complete the assigned tasks and organize regular progress meetings, participated in project implementation meetings, and completed designated documents. 2. This project assisted in the presentation of system screens at press conferences and continued to meet the needs of the authorities.

英文關鍵字

Inspection, environmental protection operation, visualization, event characteristic label, financial data analysis