Remediation
Remediation is the process of automatically fixing or improving educational content based on validation results in the Curriculum Curator system.
Overview
The remediation system in Curriculum Curator provides automated correction of content issues detected during validation. It helps improve content quality without manual intervention, streamlining the content creation pipeline.
Remediation Manager
The Remediation Manager orchestrates the remediation process by:
- Loading appropriate remediators based on configuration
- Applying remediators to content with validation issues
- Tracking changes made to the content
- Providing detailed reports on remediation actions
Remediator Types
The system includes multiple types of remediators, each addressing different aspects of content improvement:
AutoFix Remediators
- Format Corrector: Fixes formatting issues like inconsistent headers, lists, and spacing
- Sentence Splitter: Breaks long sentences into shorter, more readable ones
- Terminology Enforcer: Ensures consistent use of terminology according to glossaries
Language Remediators
- Translator: Translates content to a different language
- Style Adjuster: Modifies writing style to match target audience
Rewrite Remediators
- Rephrasing Prompter: Uses LLMs to rewrite problematic sections while preserving meaning
- Simplifier: Reduces complexity for specified reading levels
Workflow Remediators
- Flag for Review: Marks content for human review when automated fixes aren't possible
- Escalation: Routes content to specialized workflows based on specific issues
Remediation Configuration
Remediators are configured in workflow YAML files:
stages:
- name: "Auto-Remediate Content Issues"
type: "remediation"
remediators:
- name: "sentence_splitter"
parameters:
max_sentence_length: 25
preserve_technical_terms: true
- name: "format_corrector"
parameters:
fix_headers: true
fix_lists: true
- name: "rephrasing_prompter"
parameters:
severity_threshold: "high"
preserve_technical_accuracy: true
Remediation Process
When remediation is triggered:
- The system analyzes validation results to identify issues
- It selects appropriate remediators based on the issue types
- Remediators are applied in a configured order
- Changes are tracked and recorded
- Content is optionally re-validated to ensure issues are resolved
Custom Remediators
The system supports creating custom remediators by:
- Implementing the base Remediator interface
- Registering the remediator with the system
- Configuring the remediator in workflows
Integration with Validation
Remediation works closely with the validation system, using validation results to target specific issues. This creates a powerful pipeline where content can be continuously improved through iterative validation and remediation.
For more information on how validation and remediation work together, see ADR-0002: Validation and Remediation Design.
Remediation in Workflows
Remediation is typically included as a specific stage in workflows, often following validation. This ensures that content issues are systematically addressed before proceeding to further processing or delivery stages.
Human-in-the-Loop Remediation
For complex issues that can't be fully automated, the system supports human-in-the-loop remediation where:
- Automated fixes are applied where possible
- Complex issues are flagged for human review
- Suggestions are provided to guide human editors
- Manual changes are recorded for future learning