AI Scoring Methodology

Learn how our AI evaluates content for truth and reasoning quality

6 Reasoning Types
19 Logical Fallacies
5 Content Types
100 Max Score

Supported Content Types

Our AI scoring system evaluates various types of content:

PostCommentArticleTweetVideo

Scoring Components

Truth Score (0-50 points)

Evaluates the factual accuracy and credibility of claims made in the content.

Factors considered: Source credibility, factual claims accuracy, evidence quality, and verifiability of statements.

Reasoning Score (0-50 points)

Assesses the logical structure and quality of arguments presented.

Factors considered: Logical consistency, argument structure, evidence support, and reasoning clarity.

Reasoning Types & Weights

Different types of reasoning are weighted based on their logical strength:

Deductive Weight: 0.9
Strongest - Premises guarantee conclusion
Inductive Weight: 0.7
Moderate - Generalizations from observations
Abductive Weight: 0.5
Weaker - Best explanation given limited data
Analogical Weight: 0.6
Moderate - Comparisons and analogies
Causal Weight: 0.8
Strong - Clear cause-effect relationships
Statistical Weight: 0.75
Good - Data-driven analysis

Logical Fallacies & Penalties

Detected logical fallacies result in score penalties:

Strawman Penalty: -0.15
Misrepresents opposing argument
Ad Hominem Penalty: -0.2
Attacks character not argument
False Dilemma Penalty: -0.1
Presents only two options falsely
Slippery Slope Penalty: -0.1
Chain reaction without evidence
Circular Reasoning Penalty: -0.15
Conclusion assumes the premise
Appeal To Authority Penalty: -0.05
Cites irrelevant authority
Bandwagon Penalty: -0.05
Appeals to popularity
Hasty Generalization Penalty: -0.1
Broad conclusion from limited evidence
Red Herring Penalty: -0.1
Introduces irrelevant information
Post Hoc Penalty: -0.1
False cause-effect assumption
Burden Of Proof Penalty: -0.05
Shifts responsibility for evidence
Equivocation Penalty: -0.1
Uses ambiguous language misleadingly
Appeal To Emotion Penalty: -0.1
Manipulates emotions over logic
No True Scotsman Penalty: -0.05
Changes definition to avoid counterexample
Appeal To Obviousness Penalty: -0.3
False Premise Penalty: -0.4
Oversimplification Penalty: -0.3
False Dichotomy Penalty: -0.3
Non Sequitur Penalty: -0.4

Final Score Calculation

Scoring Formula:
Final Score = Truth Score (0-50) + Reasoning Score (0-50) + Fallacy Penalties
  • Truth Score: Based on factual accuracy and source credibility
  • Reasoning Score: Calculated using weighted reasoning types
  • Fallacy Penalties: Negative points for detected logical fallacies
  • Range: Final scores are clamped between 0 and 100
Note: This scoring system is continuously improved through machine learning and human feedback to provide the most accurate content evaluation possible.