Explainable Propaganda Detection via Sentence Structural Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for detecting propaganda in news articles lack explainability, making it difficult for users to trust and understand the accuracy of the results.
Innovation Solution
A method and system that analyze text data, specifically news articles, using a machine learning model to identify propaganda by parsing sentences for structural details, identifying dimensional features, and assigning covariance scores to detect propagandizing tactics, while providing explainable results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing systems detect propaganda in news articles, then propaganda identification capability is improved, but explainability and user trust are worsened
Solution Approach 1:
The system segments the propaganda detection process into multiple analytical components: sentence-level structural analysis, dimensional feature extraction, and tactic-specific covariance scoring. Each component generates intermediate results that collectively form the final explanation, allowing users to trace the detection logic through distinct analytical stages rather than receiving a single black-box result.
Solution Approach 2:
The system introduces dimensional features and covariance scores as intermediary elements between the raw text input and final propaganda classification. These intermediaries serve as explainable representations that bridge the gap between complex machine learning computations and human-understandable evidence, showing users what specific linguistic patterns triggered each detection.
2Ease of operation
If the system provides detailed propaganda detection results, then user understanding is improved, but system complexity is worsened
Solution Approach 1:
The complex analysis is divided into manageable segments that process different aspects of propaganda detection independently: structural parsing, feature extraction, and tactic classification. This modular segmentation allows the system to maintain high analytical capability while presenting results in organized, user-friendly categories that reduce perceived complexity.
Solution Approach 2:
The system transforms complex multi-dimensional feature spaces into simplified visual representations by projecting high-dimensional covariance scores onto interpretable tactic categories. This dimensional transformation enables detailed analysis to be communicated through straightforward labels and scores that users can comprehend without needing to understand the underlying complex feature interactions.
Data Source
AI summary
Explainable propaganda detection is provided by parsing each sentence of a plurality of sentences in an article for structural details; identifying, via a machine learning model, dimensional features across the plurality of sentences based on the structural details; identifying, via the machine learning model, a propagandizing tactic demonstrated in each sentence of the plurality of sentences based a covariance score to the propagandizing tactic and the plurality of dimensional features identified for a given sentence; and displaying, in a user interface that includes the article, propaganda labels corresponding to the propagandizing tactic identified for each sentence of the plurality of sentences.