Understanding Bias & Sentiment Ratings

1. How Lamu News Analyzes Bias

Lamu News utilizes advanced language models to evaluate how different media outlets report on the same event. Rather than declaring a single outlet as absolutely unbiased, we estimate how a story is framed politically. This gives readers the tools to understand the perspective of the source article.

Our system extracts the core facts of an article, analyzes the loaded vocabulary, and compares the framing against standard political spectrum definitions.

2. Left, Center, and Right Percentages

Every article details a political framing breakdown shown as percentages for Left, Center, and Right. These values always sum up to 100%:

Left Percentage: Reflects framing, vocabulary, or source usage typically associated with progressive or left-leaning political perspectives.

Center Percentage: Reflects neutral, objective, or bipartisan reporting that avoids partisan framing.

Right Percentage: Reflects framing, vocabulary, or source usage typically associated with conservative or right-leaning political perspectives.

3. The Mathematical Bias Score

To provide a single tracking metric, we derive a political bias score ranging from -1.00 (maximum left framing) to +1.00 (maximum right framing).

The formula is simple and transparent:

Bias Score = (Right % - Left %) / 100

For example, if an article has 60% Left, 30% Center, and 10% Right framing, its derived bias score is (10 - 60) / 100 = -0.50, which points to a moderate Left bias label.

4. Sentiment Scores and Labels

Sentiment analysis measures the emotional tone of the writing. The sentiment score ranges from -1 (extremely negative or critical) to +1 (extremely positive or laudatory), with 0 representing neutral reporting.

We group articles into three sentiment labels: Positive, Neutral, or Negative. Most high-quality objective reporting tends to fall near 0 (Neutral).

5. AI Estimates and Disclaimer

It is important to remember that these metrics are strictly AI-estimated evaluations generated by computational language models. They do not represent objective, absolute truth or established facts. AI evaluations are designed as supplementary tools to encourage media literacy and critical reading.