/
AI Map Training Guide: An In-Depth Overview

AI Map Training Guide: An In-Depth Overview

Owned by Rajeev Rai
Last updated: Nov 25, 2024
3 min read

Introduction to AI Map

AI Map is an advanced tool for automating the process of data mapping between source and target systems. It leverages AI capabilities to suggest mappings with varying confidence levels—High, Medium, and Low—and integrates with libraries to save and reuse mappings for efficiency. Designed for flexibility and accuracy, AI Map minimizes manual intervention and accelerates data integration tasks.

1. Setting Up Your AI Map Environment

Before starting:

  • Ensure the required automation processes (e.g., Purchase Order Automation) are installed.

  • Open a blank mapping layout or an existing one where mappings can be applied or edited.

 

2. Features of AI Map

A. Auto Mapping

  • Automatically maps fields between source and target layouts when there is a high level of similarity.

  • Ideal for quick and accurate results when layouts are identical.

B. Manual Mapping

  • For cases where auto mapping doesn’t suffice, users can manually create mappings between fields.

  • Manual mappings can be saved to the AI Map Library for future reuse.

C. AI Map Library

  • Stores manual mappings for future use.

  • When similar layouts are encountered, saved mappings are presented as High Confidence suggestions.

D. Confidence Levels in Mapping

  1. High Confidence:

    • Automatically generated for exact matches or saved library mappings.

  2. Medium Confidence:

    • Suggested for partial matches or when fields are somewhat similar.

  3. Low Confidence:

    • Generated by AI (via Gen AI services like OpenAI) when confidence is low.

    • Requires manual review and approval before use.

3. Applying Mappings

Using Confidence Filters

  • Users can choose to apply mappings based on confidence levels.

  • If data accuracy is critical, only High Confidence mappings should be applied.

  • Medium and Low Confidence mappings can be manually reviewed, adjusted, or rejected.

Modifying and Reviewing Mappings

  • Adjust the mapped fields as needed.

  • Reject or remove unwanted suggestions without affecting the rest of the mappings.

  • Use the review mechanism during runtime to verify that AI-generated suggestions meet system requirements.

4. Advanced Mapping Scenarios

  • AI Map recognizes different naming conventions and intelligently maps fields with equivalent meanings (e.g., “First Name” and “F Name”).

  • In complex cases, users can create detailed mappings that are added to the library for future efficiency.

Custom Rules

  • AI Map is evolving to allow natural language-based rule definitions, enabling automation of more intricate mappings. For instance, users may soon be able to instruct the system to "convert date formats" or "truncate values" without manual intervention.

5. AI Map in Action: Design Time vs. Runtime

  • Design Time:

    • All mappings are created, tested, and finalized during this phase.

    • Ensures that rules and mappings are fixed before deployment.

  • Runtime:

    • AI Map-generated data is subject to user review before passing to target systems.

    • This prevents errors and ensures data reliability.

6. Common Issues and Error Handling

  • Missing Mappings:

    • AI Map doesn’t currently alert users about unmapped required fields, but manual checks are recommended.

    • Suggest adding this feature in future updates for better usability.

  • Review Requirements:

    • Always review AI-suggested data before applying, especially for low-confidence mappings.

    • Avoid bypassing the review step unless confident in the system's accuracy over time.

7. Planned Features for Enhanced Usability

  • Automating complex rules through plain-language commands.

  • Improving AI-driven suggestions for Medium and High Confidence mappings.

  • Introducing warnings for unmapped mandatory fields during the design phase.

8. Best Practices for AI Map

  1. Start with Auto Mapping to minimize effort.

  2. Save manual mappings to the library to improve future mapping efficiency.

  3. Prioritize reviewing low-confidence mappings for accuracy.

  4. Use the design-time review process to ensure mappings are tested and verified before runtime.

Conclusion

AI Map is a powerful and evolving tool that streamlines data mapping processes while maintaining flexibility and accuracy. By leveraging its automation features, library capabilities, and AI-driven suggestions, users can save time, reduce errors, and handle complex integrations with ease.