Introduction to the Use Case Structure

Aralia Use Case: SDGs Inclusive Growth

Aralia Use Case: Public Works Bid Management and Risk Alert System

Aralia Use Case: Cabbage Supply and Demand Imbalance Analysis

🎯 1. Problem Statement & Background

• Description of the Problem: Public Works Bid Management and Risk Alert System
• Context Statement:
  • The amount of government procurement is huge (NT$ 800 billion per year, accounting for nearly 40% of Taiwan's civil engineering projects).
  • However, bidding information is scattered and checking resources are limited, resulting in "lack of consistency in selection" and "information asymmetry".
  • Small and medium-sized enterprises (SMEs) are at a disadvantage in terms of bidding information, which often degenerates into a price war.
• Stakeholder/User Persona
  • Government procurement units: want to effectively screen vendors and reduce the risk of failure.
  • Bidding vendors: need more fair and transparent competition information.
• Possible Questions:
  • Which vendors have poor performance in the past?
  • Is there a potential risk in a project bid?

📊 2. Data Sources

• Primary Sources:

  • Civil Engineering Planet:
    • Government Procurement Website: Tenders, Bid Decisions, Contract Compliance, Record Checks
    • Public Works Commission: Principles of Inspection, Construction Inspection Data
    • Occupational Disaster Data Planet: Records of major occupational disasters.
  • Financial Planet: Balance Sheet, Cash Flow Statement

  Data Name	Source	Description	Remarks
  Tender Award Information	Civil Engineering Private Planet	Information on the results of public works tender decisions, including tender, decision, date, amount, etc.
  Project history	Civil Engineering Private Planet	Vendor's name, contract history, amount, and quantity.
  Occupational safety records	Civil Engineering Private Planet	Vendor name, date, type of disaster, number of injuries, owner, project name
  XBRL	Financial Reports Planet	Quarterly and annual balance sheets of listed companies

• Data Type:

  • Structured Forms (Case Number, Vendor Name, Tender Amount...)
  • Ratings and performance records (comprehensive engineering ratings)

• Pre-processing:

  • Integration of different source fields
  • Standardization of vendor names
  • Completion of missing values and formatting consistency

🛠️ 3. Aralia Demo

• User's Operation Workflow:

  • Import government procurement data → create "Civil Engineering Private Planet".
  • Create visualization charts: Bid volume trend, bid award amount distribution
  • Save Landmarks: Bid risk monitoring dashboard

• Visualization Examples:

  • Trend of Bid Amount by Year (line chart)

  

  • Vendor Bid Rate Comparison (bar chart)

  

• Landmark Preview: Chart will be refreshed when data is updated.

📈 4. Analytic Results and Insights

• Results in Brief:
  • Certain vendors have experienced multiple delays or workplace disasters over the past five years.
  • There is a high correlation between abnormally low bid amounts and delays in contract fulfillment.
• Insight Highlights:
  • Vendor management capabilities can be quantified in terms of "workplace disasters" and "performance on contract".
  • Comprehensive engineering rating performance more accurately reflects supplier risk
• Graphical Illustration: Engineering Rating Performance Comparison Chart

• Recommended Actions:
  • The government can incorporate risk indicators into bid evaluation.
  • Vendors can self-assess their competitive strengths and weaknesses and adjust their bidding strategies.

🧠 5. AI Agent Demo

• Sample User Question:

  • "Please list the vendors with more than 3 delay records in the past 5 years".

  

  • "Which recent bids were significantly lower than the reserve price?"

• Sample Agent Response:

  • Natural language → automatically query the Data Planet → reply with a table and explanation.
  • Provide decision basis, e.g., "Vendor A has a 45% default rate, which is high risk.”

🔧 6. Agent Setup Tutorial

• Steps:
  1. Create a Data Planet (bid procurement and fulfillment data is currently a private planet, one must create their own Planet)
  2. Connect to OpenRAG™ framework
  3. Build inspection AI Agent
  4. Integrate Chat Widget on the front-end to support natural language queries.
• Flowchart:
  • Data Planet → OpenRAG → Agent → User Q&A
• Common Errors:
  • Vendor name inconsistency → create an alias comparison table.
  • Data update delay → set up program to refresh automatically.

🧾 7. Conclusion & Next Steps

• Extended Applications
  • Procurement by local counties and municipalities.
  • Other bid analysis in transportation and construction fields.
  • Risk assessment of civil engineering and large-scale construction.
• We Encourage Users to:
• Call to Action (CTA):