CDF File Annotation - Streamlit Dashboard UI/UX Improvements to Handle Pattern Mode

[lucasguimaraes-rdx]

CDF File Annotation – Streamlit Dashboard UI/UX Improvements for Pattern Mode

Overview

This document outlines a set of planned improvements to the Streamlit Dashboard in Pattern Mode, focusing on enhancing user experience (UI/UX), improving data clarity and increasing interaction efficiency with annotation promotion and pattern creation.

These changes aim to make the interface more intuitive, dynamic and consistent with the new integrations coming from pattern mode.

---

Planned Improvements

Filtering and Metadata Enhancements

• Replace “Filter by Resource Type” with “Filter Quality Table”, allowing users to filter by any metadata property.
• Apply the selected metadata property filter across all relevant components, including:
  • “Annotation Coverage Percentage”
  • “Annotation Coverage by Resource Type” chart
  • “Actual Annotations vs. Potential New Annotations” table
  • "Pattern Catalog"
• Rename “Resource Type” to match the field name provided by the extraction pipeline (secondaryScopeProperty).
• Replace the hard-coded sysUnit field with a dynamic secondaryScopeProperty sourced from the extraction pipeline.

---

Annotation Type Logic

• Consider pattern edge status when classifying annotations:
  • Suggested → Potential Annotation
  • Approved → Actual Annotation
• Include approved pattern annotations in the total Actual Annotations, which means now they are an union of tags/docs annotation tags and approved pattern annotation tags.
• Incorporate edge status when regenerating canvas objects.
• Add a new field in the Pattern Catalog to distinguish between manual and automatic patterns.
• Capture which user promoted an edge.
• Replace the asset resource type “equipCategory” with “equipType”.

---

Visualization and Interaction Improvements

• Split the “Annotation Coverage by Resource Type” chart into two separate charts:
  • One for assets
  • One for files
• Display both name and externalId in file references for better recognizability.
• Add an “Associated Files” expander in the “Tag to Connect” section to surface related files without cluttering the UI.
• Convert manual text inputs into dropdown fields when creating patterns.
• Add loading screens while data is being fetched.
• Replace generic success messages with toast notifications, especially when:
  • Adding manual patterns
  • Regenerating the canvas
• Add a Reset button to clear cached data.

---

Performance and Architecture

• Optimize performance by balancing between multiple dataframes versus shared dataframes

---

Support for Auto-Promote Pattern Edge

• Update the UI to display new edge statuses depending on the tags property:
  • Rejected -> no matching tag found, so it wasn't promoted
  • Ambiguous -> multiple possible matches, so it wasn't promoted
• Incorporate the new statuses into the "Actual Annotations vs. Potential New Annotations" table
• Implement a flag that allows a cache mechanism for manual promotions in the Overall tab to support auto-promote

---

Goal

These improvements aim to deliver a smoother experience for users working with the annotation dashboard, ensuring stronger consistency with the Pattern Mode integration