Research Methodology
News Literacy employs a systematic approach to news analysis, combining automated data collection with advanced natural language processing to produce objective news summaries and identify patterns of editorial framing across the political spectrum.
1. Source Selection
Our source registry includes 50+ news outlets spanning political, regional, and cultural dimensions — from left-leaning publications to centrist wire services, right-leaning media, and international outlets across multiple regions. Sources are categorized using the Ad Fontes Media Bias Chart as a reference framework.
Political Leaning Scale
Each source is assigned a leaning score from -5 (far left) to +5 (far right), with 0 representing centrist or wire service reporting. This scoring enables quantitative analysis of coverage patterns.
2. Article Collection
The platform performs a daily automated collection of articles from all registered sources. This process operates with full respect for each publication's terms of service and robots.txt directives.
- RSS feeds are parsed to identify new articles
- Full article text is extracted using specialized parsing algorithms
- Rate limiting ensures responsible data collection (max 2 requests/second per domain)
- Articles are timestamped and attributed to their original source
3. Story Clustering
Articles covering the same news event are grouped together using natural language processing techniques. This allows comparison of how different sources report on identical events.
TF-IDF Vectorization
Article headlines and opening paragraphs are converted to numerical vectors using Term Frequency-Inverse Document Frequency (TF-IDF) analysis. Agglomerative clustering then groups articles with high cosine similarity.
The top 50 stories are selected based on cross-source coverage—stories covered by more sources across the political spectrum receive priority.
4. Fact Sheet Generation
For each story cluster, an objective fact sheet is synthesized that includes only verifiable information: confirmed facts, direct quotes with attribution, and documentation of where sources disagree on factual claims.
Fact Sheet Components
- Neutral Headline
- A headline stripped of loaded language or editorial framing
- Objective Summary
- A narrative composed only of verified, attributed facts
- Key Facts
- Bulleted list of confirmed information with source attribution
- Confirmed Quotes
- Direct quotes from primary sources with context
- Unresolved Claims
- Factual assertions that conflict between sources
5. Bias Analysis
Each source's coverage is analyzed for editorial framing patterns. This analysis identifies specific language choices, emphasized facts, and notable omissions that reveal how framing shapes reader perception.
Analysis Dimensions
- Framing: The editorial angle or narrative frame applied to the story
- Language: Specific word choices that carry emotional or political weight
- Emphasis: Which facts receive prominent placement
- Omission: Notable facts covered by other sources but absent here
- Tone: Overall emotional register (measured, alarming, celebratory, etc.)
6. Limitations & Transparency
This platform is designed as a research and educational tool. Users should understand its limitations:
- Analysis is performed once daily and may not reflect breaking developments
- Source selection, while diverse, cannot include all perspectives
- Automated analysis may occasionally misinterpret nuanced language
- Political leaning scores are approximations based on aggregate assessments
- Original reporting should always be consulted for complete context
Links to all original articles are provided so readers can evaluate sources directly.
7. Research Applications
News Literacy is designed to support several research and educational objectives:
- Media literacy education and critical reading skills development
- Longitudinal studies of coverage patterns across the political spectrum
- Analysis of how framing affects public perception of events
- Development of automated bias detection methodologies