Taipei City's Geotechnical Engineering Office Launches AI-Powered SCWQP Review System
- APMIC
- May 6
- 2 min read
Challenges in the Traditional Review Process
Before the implementation of the AI system, the Geotechnical Engineering Office faced several persistent issues:
Lengthy Review Timelines: Scattered data across different formats led to low audit efficiency and prolonged review cycles.
Human Error: Manual verification of data often resulted in omissions and inaccuracies, undermining the integrity of the reviews.
Limited Feedback Mechanisms: A lack of immediate feedback meant that revision cycles were unnecessarily long and inefficient.
Resource Constraints: The heavy workload on limited staff strained administrative capacity, reducing overall efficiency.
These challenges underscored the urgent need for a streamlined and automated solution to improve the reviewing process.
Transforming the Process with AI
To overcome these challenges, the Geotechnical Engineering Office collaborated with APMIC to develop an AI-powered SCWQP Review System. This system offers several transformative benefits:
Accelerating the Review Process: By digitalizing graphic files and automating data retrieval and analysis, the system significantly enhances review efficiency.
Reducing Human Omissions: Automated data comparison and verification ensure completeness and minimize errors.
Providing Instant Feedback: The system generates automated recommendations for compliance with regulatory standards, shortening revision cycles.
Enhancing Administrative Efficiency: By optimizing workflows and reducing labor costs, the system alleviates the burden on administrative staff.
Development of the AI SCWQP Review System
1. AI-Assisted Pre-Screening Model
The system utilizes machine learning and optical character recognition (OCR) to convert text images from original documents into readable and analyzable data. Natural language processing (NLP) further enables the AI to comprehend document content, assess compliance, and identify abnormalities. At APMIC, advanced models like BERT and GPT are used to enhance text classification and comprehension, ensuring high accuracy in identifying errors and anomalies.
2. API Development and Cloud Deployment
To integrate the AI pre-screening system seamlessly into the soil and water conservation management platform, APMIC developed APIs for automatic transmission and feedback of results. Cloud deployment ensures high performance and stability, enabling large-scale application of AI in real-world review scenarios. The cloud infrastructure also supports dynamic monitoring and performance optimization for continuous improvement.
How Is It Different from General AI Tools Like ChatGPT?
APMIC offers tailored solutions through a cloud-based AI training platform and on-site deployment, providing customized technology and hardware configurations to meet specific organizational needs and budgets. As an NVIDIA-certified ecosystem partner with years of experience in large language models (LLMs), APMIC has successfully assisted over 1,100 companies in adopting AI solutions. Additionally, with ISO 27001 certification, data security is assured throughout the process.
Results of the AI SCWQP Review System
During the testing phase, the AI system processed over 500 applications, demonstrating significant improvements in accuracy, performance, and stability. Review times per application were drastically reduced, and the system operated without interruptions, ensuring reliability in practical scenarios. By enabling data exchange and visualizing pre-screening results, the system provides users with an intuitive interface, enhancing review efficiency and reducing the need for returns and resubmissions. This achievement marks a significant milestone in achieving intelligent, automated reviews.
コメント