This dashboard presents a comprehensive analysis of online video content, utilizing multiple evaluation metrics to assess quality and compliance with content standards. The analysis is available in two viewing modes:
• Aggregated View: Displays trends and patterns across all analyzed videos, providing insights through interactive visualizations and statistical summaries.
• Raw Data View: Shows individual video assessments with detailed justifications and supporting examples for each evaluation metric.
Each video is evaluated across multiple dimensions including:
- Sentiment and Tone Analysis
- Bias and Fairness Assessment
- Factual Accuracy Verification
- Plagiarism Detection
- Ethical Considerations
- Content Objectivity
- Transparency in Presentation
Evaluations are supported by specific examples extracted from the content, providing context for each assessment. The analysis uses a combination of automated tools and expert review to ensure accuracy and consistency in the evaluation process.
Toggle between views using the mode selector above to explore different aspects of the analysis. Use the search and filter options to focus on specific content or metrics of interest.
Methodology
This analysis employs a two-stage AI processing pipeline:
Stage 1: Deep Analysis
Using a locally hosted DeepSeek Reasoning R1:14B model, each video transcript undergoes detailed analysis through Chain of Thought (COT) reasoning. This model excels at:
- Complex content evaluation
- Multi-step reasoning
- Contextual understanding
- Example identification
Stage 2: Structured Output
The COT results are processed by GPT-4o-Mini to:
- Convert reasoning into structured JSON
- Extract quantifiable metrics
- Organize supporting examples
- Generate consistent scoring
Analysis Metrics & Prompts
Sentiment Analysis
Evaluates tone, word choice, and context to determine emotional direction.
"Analyze the sentiment focusing on tone, word choice, and context."
Bias & Fairness
Detects linguistic bias, stereotypes, and unfair framing.
"Analyze for stereotypes, loaded language, unsupported generalizations, or unfair framing."
Factual Accuracy
Verifies claims against reputable sources.
"Verify claims against reputable sources."
Plagiarism
Identifies unattributed content from published sources.
"Identify content mirroring published sources without attribution."
Ethical Concerns
Flags content violating ethical standards or containing sensitive topics.
"Flag ethical violations and sensitive content. "
Clarity
Evaluates logical flow, structure, and readability.
"Assess logical flow, structure, grammar, and readability. "
Objectivity
Evaluates balance of perspectives and evidence presentation.
"Evaluate perspective balance and evidence presentation. "
Transparency
Checks for disclosure of conflicts, sponsorships, and sources.
"Check disclosure of conflicts, sponsorships, and sourcing. "
Accountability
Identifies grammatical, factual, or logical errors.
"Flag grammatical, factual, or logical errors. "