The Seekr Score
Seekr is a news site and search engine that uses proprietary Artificial Intelligence (AI) technology to index thousands of websites and top news sources using natural language processing.
Seekr takes the given query and applies a suite of machine-learning algorithms, evaluating all results with the same scrutiny as a data scientist and expert journalist. Any time our AI detects things like persuasive techniques, contradictions, or clickbait etc. it impacts the overall score. Once complete, the Seekr Score algorithms generate a mathematically based rating of the content so people can make more informed choices about what they read before they click. For details, check out the list of factors that our AI is trained to detect and score.
- The Seekr Score provides information transparency, making it easy for users to discern quality before choosing to click. The Seekr Score will eventually expand beyond news to other relevant categories and will be applied to publishers, authors, journalists, etc.
- The Seekr Score Filter gives you the ability to filter information giving you choice and control.
A numerical scale from 0 to 10, where ten indicates the highest quality and zero the lowest one.
- Scores equal or higher than 6.7 show a high-quality article that is likely to be informative.
Those pieces are written objectively and may include accurate headlines, a variety of sources, subject matter expertise, etc. - Mid-range scores, from 3.4 to 6.6, indicate an increased detection of inaccuracies. Any claims or
conclusions should be verified by additional sources and cross-checked against higher scoring articles. - Articles with a score equal or lower than 3.3 will indicate a high level of inaccuracies and higher use
of persuasive techniques, contradictions, or attempts to mislead that are inconsistent with journalistic
principles and reputable news writing.
Clicking on the score will give you an in-depth explanation, including the factors that contributed to the score and descriptions of those factors.
Seekr evaluation algorithms offer a consistent and programmatic assessment of results in just seconds. For each result, Seekr AI analyzes the quality of information, adherence to journalistic principles, and general reliability of the author and domain source.
Each article is analyzed using several machine-learning algorithms to detect the presence of specific criteria that can indicate techniques used to mislead (i.e., subjectivity, absence of author, hyperbole, clickbait, etc.) Once analyzed, the result is scored using an aggregate numerical rating based on several sub-features that contribute to the overall score. Each feature of analysis answers key questions about the quality level of the article.
It is designed to help users evaluate and compare information supporting more informed and confident opinions.
The table below explains the features present in our algorithms.
| Feature | Description | Reasoning |
|---|---|---|
| Ad Hominem | Ad hominem arguments attack the character, motive, or other characteristics of a person rather than their actual argument. This can also be summarized as a personal attack. | Any credible article, whether news or opinion, should focus on evidence and arguments rather than attacking personal characteristics. |
| Byline Absence | A byline is a line in an article that names the author(s). | Absence of a named author indicates a lack of accountability. |
| Clickbait | Clickbait headlines use hyperbolic (and often misrepresentative) language to appeal to curiosity and sentiment, teasing the reader to click rather than informing them about the topic of the article in a non-sensationalist manner. | Credible headlines should prioritize reporting the truth as accurately as possible rather than using them to drive clicks. |
| Incoherence | Incoherence is the disagreement between the headline and body copy or contradictions within the story as well. | Inconsistencies can be a result of multiple writers or editors with bias seeping in, or simply a poorly researched and supported thesis. Often tied to clickbait. |
| Subjectivity | Subjective writing includes the author’s opinions and feelings. Objective writing reports factual, undisputable information. | Credible news articles are expected to be reported objectively, meaning they have low subjectivity. Only opinion pieces (i.e. op-ed) are expected to be subjective. |
| Title Exaggeration | An exaggerated headline portrays something as more important than it is in reality or hyperbolizes the story findings to entice the reader to click. | Exaggeration distorts the truth and can mislead. Often tied to clickbait. |
News domains are transparent when they share truthful information about themselves such as their mission, ethics, ownership, and policies. Transparency is key in building trust, and fully transparent domains are likely to be more reliable.
The Overall Domain Transparency Score Includes:
- Ethics
- Fact-Checking
- Types of Work
- Author Information
- Mission
- Corrections
- Ownership
- Newsroom Contact
- Diversity
- Sources
- Leadership
- Policy Accessibility
Seekr strives to be as accurate as possible. We are always working hard to improve our algorithms.
Some areas we are actively working to improve, and build are:
- Non-English websites are not yet rated
- We may not be able to rate some sites/articles behind a subscription wall
- We do not factor tweets, videos, and photos into the analysis
