The primary nine tasks of platform establishment in this project are as follows: 1. Maintain the platform functions (1) Main System This project optimized platform user-friendliness according to user feedback. The improvements, including batch inquiry of enterprise information, case modification and review system, WEB interface of smart logic information, and value-added services of other data, were made to enhance information demonstration. (2) GIS System The GIS system was improved based on user feedback. To cope with the inspection and management of temporarily registered factories after the Factory Management Act came into effect, the inquiry information and map layers of temporary factories were updated on a rolling basis. This assisted inspectors in understanding the status of temporary factories and providing a map of the new agricultural irrigation channel system immediately, strengthening the overall effectiveness of inspection. (3) Sub-system This year, the app was updated twice before October to resolve the long-standing problems. The updates included the addition of schedule functions, which allowed inspectors to directly make the schedule through tablets. Three sessions of educational training were organized in October. During the training, the users posed fewer questions than in previous training sessions. This indicated that the current app could meet the inspection requirements. 2. Ensure the operation of data interface The platform has completed data collection from 16 business units. All the data have been obtained through automatic data interface, with a total of 386 datasheets. 3. Research and analyze the logic of potential behavior Through the professional knowledge, logic of thought, and years of experience of senior inspectors as well as the advanced analysis conducted by computers, this project established various examples of different types of inspector cases to pass on the experiences of onsite inspectors and improve the inspection efficiency. The analysis module was established based on the existing business permit and reporting data so that the inspectors could quickly find the abnormal conditions of the business and automatically import the analysis results into the sub-system (app). These functions enabled the inspectors to identify abnormalities immediately and conduct related inspections. 4. Research and analyze the Sheet of Potential Environmental Load of an enterprise The inspectors could quickly examine the key information on the Sheet of Potential Environmental Load of an enterprise. Information, including basic information, abnormal data, potential environmental load, and necessary graphs, was provided to the inspector to quickly review and understand the current status. 5. Research and analyze the relevance of specific inspection cases By using the waste stream flow, relevant information, such as waste receipt, production, clearance, waste category, quantity, and clearance objects, was indicated on a graph. This design could quickly recognize if an inconsistency occurred between the reporting quantity from up- and downstream enterprises. 6. Research and analyze the inferential analysis model of violation rate In the past four years, the audit logic and analysis of abnormal enterprise data were established based on enterprise basic information and reporting data. However, some operational problems could not be easily identified. With the development of big data applications, illegal behavior could be predicted by using inferential statistics models, so the data analysis method this year changed from narrative statistics to inferential statistics. This project examined past violation cases and found that the predictive model of “operational issues” was highly accurate. The model could effectively predict the occurrence of violation in operation by using independent variables. However, compared to inferential statistics, analyzing “permitting issues” and “reporting issues” by using narrative statistics generated more accurate results. The accuracy could reach greater than 80%. 7. Adjust the graphical presentation of the platform The working method of inspection topics was presented intuitively and visually. For example, the working scope of the three Environmental Chief Inspection Teams was shown on the platform. Specific data labels were designed to be selected to demonstrate related information to broaden the inspection range. 8. Research and analyze the GIS system platform to show the area model of enterprise potential environment load In the past four years, the GIS platform was developed to display the geographic information of enterprises in graphics. The GIS system also provided spatial information such as enterprise information, buffer analysis, source tracing inquiry, and device positioning. In 2021, the function of “areas of potential environmental load” will be introduced into the GIS platform for inspectors to identify suspicious objects intuitively. This will assist inspectors in reducing the preparation time. 9. Research and analyze the voice function of the pilot version of inspector assistant (Grace) This project used the semantic parsing tool supported by Actions on google to gradually establish a voice prototype by integrating Dialogflow on the server. The system categorized enterprises by asking for registration number or company name, followed by semantic analysis, semantic modeling, and semantic training. Finally, the information was introduced into three query functions (e.g., function operation, pollution regulations, and enterprise information) and one work assignment service (pre-scheduling). 10. Organize training sessions on platform functions and data analysis This year, in response to the pandemic, the training was conducted by video, focusing on topics of Enhancement of Existing System Functions and Their Operation Methods and Newly Developed System Functions. The training explained topics, such as the overall structure of the abnormal logic, methods of accessing enterprise information and abnormal data, and interpretation of the abnormal patterns, to effectively improve system operation. 11. Coordinate and compile customer feedback and communication management During the project implementation period, the project team consistently provided users with quality and various service channels for feedback or consultation. This project continued to carry out administrative work, including routine work meetings and other administrative tasks.

Inspection, environmental protection business, information access, smart logic