About This Project

RageBait Score is a tool designed by me, a single developer, to bring awareness to emotionally charged content in news media. By analyzing headlines, it identifies tones that may influence readers' emotions, aiming to promote media literacy and critical consumption of news. Yes, I understand the hypocrisy of having a tool that tries to identify sensationalism called RageBait.

Glossary

• RageBait Score: A metric indicating the proportion of emotionally charged headlines on a site.
• Most Frequent Tone: The tone category most commonly detected in a site's headlines.
• Match Rate: The percentage of headlines that matched any tone category.
• Tags: Labels assigned to headlines based on detected emotional tones, such as 'Fearful' or 'Optimistic'.

Tone Categories

• Fearful: Content that evokes fear or anxiety.
• Angry: Content that incites anger or outrage.
• Misleading: Content that presents information in a way that distorts, omits, or exaggerates facts to influence perception.
• Clickbait: Sensationalized content designed to attract attention and clicks.
• Polarizing: Content that divides opinion or emphasizes conflict.
• Optimistic: Content that conveys hope or positivity.
• Research: Content that presents facts or analysis such as research and studies.

How Scoring Works

RageBait Score utilizes a combination of a keyword engine and sentiment analysis to detect emotional tones in headlines. The process involves:

• Keyword Matching: Headlines are scanned for specific keywords associated with each tone category.
• VADER Verification: The VADER Sentiment Analysis tool [VADER GitHub] assesses the sentiment to confirm the detected tone.
• Negative matching: Matches are checked for disqualifying keywords such as commercial terms (deals and offers) as well as tragic terms (natural disasters, tragedies, etc) as those can be incorrectly matched by the sentiment analysis.

Negative tones (e.g., Fearful, Angry) are weighted more heavily in the scoring, while positive tones (e.g., Optimistic, Informative) lower the overall score. A lower match rate indicates a higher proportion of neutral content, resulting in a lower RageBait Score.

Scoring Scale

The scale is 0-100 and reflects the amount of emotional tones detected.

Limitations

With the combination of the keyword engine and sentiment analysis, tone matching is highly accurate. However, some headlines may continue to go unmatched/mismatched due to the nuance or context of the headline.

Only the homepage is scanned. The goal is to measure the emotional tone of a news site home.

To keep costs down, sites scans are cached for 4 hours.

There are plans to improve both of these limitations but as a single developer supporting this tool, support is needed. If you find this tool valuable to increase news comsumption literacy, consider supporting RageBait.

Supported Sites

The site is built to scan news related websites. It parses through anchors (links) on the homepage. Global websites elements such as header, footers and navigations are excluded. Sites that use JS loading techniques may not be supported.

Developer Resources

RageBait Score aims to be developer friendly. Data such as scans, keyword categories and headline matches can be accessed:

• /data/json: Access scan results in JSON format.
• /tags: View the list of keywords used for tone detection.
• /examples/[site]: See recent matched examples for a specific site.

For more information and documentation, visit the Documentation Page.

Disclaimer

RageBait Score applies its analysis uniformly across all scanned sites, without any preferential treatment or bias. While the combination of keyword matching and AI verification aims to ensure accuracy, false positives may occur. However, extensive testing has shown a very low rate of such instances.