As a data analyst, I want the system to objectively assess the political orientation of a snippet, so that I can understand its bias on a spectrum from -1.0 (extremely left-leaning) to +1.0 (extremely right-leaning).
Acceptance Criteria:
System Instructions Update:
The analysis prompt should now include political content analysis as a core capability.
It must emphasize objectivity, evidence-based analysis, and maintain a focus on cultural sensitivity.
Schema Updates:
JSON output should now include a political_leaning field.
political_leaning should be a float ranging from -1.0 to +1.0.
Ensure bilingual explanations with evidence citations are included.
Validate JSON schema to incorporate the new political_leaning field.
Analysis Requirements:
The system should analyze political orientation focusing on:
Explicit policy positions
Specific arguments
Language and rhetoric
Sources cited
Solutions proposed
Scoring should be evidence-based with references to the snippet content.
Explanations must include specific citations.
Example Outputs:
Include examples demonstrating analysis across the political spectrum:
A right-leaning example (e.g., vaccine discussion)
A left-leaning example (e.g., economic policy)
A neutral example (e.g., infrastructure)
Each example must demonstrate:
Evidence-based analysis
Direct quote citations
Objective assessment
Correct schema formatting
Implementation:
Update Stage_3_system_instruction.md to reflect new capabilities.
Modify Stage_3_output_schema.json to include political_leaning.
Ensure the main instruction document is updated accordingly.
Tasks:
[ ] Update Stage_3_system_instruction.md with new political analysis instructions.
[ ] Modify Stage_3_output_schema.json to add the political_leaning field.
[ ] Update the main analysis prompt document with new requirements.
[ ] Implement validation patterns for the updated JSON schema.
[ ] Test with various examples to ensure accurate political leaning assessment.
[ ] Document any changes to the process for capturing/writing the political_leaning data field in the database.
Notes:
The prompt has been updated directly on GitHub here.
Ensure all changes are backward compatible with existing systems.
The explanation format should be clear and require specific citations for transparency.
As a data analyst, I want the system to objectively assess the political orientation of a snippet, so that I can understand its bias on a spectrum from -1.0 (extremely left-leaning) to +1.0 (extremely right-leaning).
Acceptance Criteria:
political_leaning
field.political_leaning
should be a float ranging from -1.0 to +1.0.political_leaning
field.Stage_3_system_instruction.md
to reflect new capabilities.Stage_3_output_schema.json
to includepolitical_leaning
.Tasks:
Stage_3_system_instruction.md
with new political analysis instructions.Stage_3_output_schema.json
to add thepolitical_leaning
field.political_leaning
data field in the database.Notes: