cropped-TrustServista_logo_middle

What is TrustServista?

TrustServista is a unique software platform that performs automated verification of multilingual articles, blogs and open source content, with the purpose of determining the origin of information (Patient Zero) and its trustworthiness, by relying on a mix of statistical and rule-based Natural Language Processing algorithms.

Who is it for?

TrustServista is designed to be used by analysts, content producers, content distributors who ingest and use multilingual open source information, on a global level

 Key Capabilities:

 TrustServista uses advanced Artificial Intelligence algorithms to shorten investigation times for media professionals, analysts, and content consumers.
     
       Trustworthiness Algorithm

TrustServista determines the trustworthiness of news articles using Artificial Intelligence. The trustworthiness score takes in account the article’s content, and also what sources it references back to, as well as which information it used and from where that information originated.

The trustworthiness metric is calculated automatically and should not be confused to the “truthness” of the information. Trustworthiness relates to three key areas of investigation and analysis that can be fully automated with NLP algorithms. The TrustLevel takes in account elements that look into the “who” and “how”, while Patient Zero relates more the “origin & context”.

News Content Lifecycle Coverage

TrustServista is designed for media professionals, content distribution channels and news readers alike. By covering the full news content lifecycle, from creation to consumption, TrustServista can be used to tackle misinformation and fake news propagation in a more efficient way.

Route to “Patient Zero”

TrustServista can find the original source of information used to create news content, by intelligently following implicit or explicit references combined with its unique semantic algorithm. “Patient Zero” is where a story originated: social media or blog posts, website pages or news outlets.

Patient Zero is determined by analyzing the Article Graph, obtained from linking the article to other articles through:

– Explicit links: hyperlinks contained in the article body

– Implicit links: references to other publishers, with no hyperlink

– Semantic similarities to other articles

Easy-to-Integrate SaaS Model

TrustServista is delivered as Sofware as a Service (SaaS) making it easy to adopt and use. It also offers REST APIs for integration into 3rd party software systems, such as newsroom technologies, social media platforms or customer-centric widgets and web browser add-ons.

TrustServista API

The TrustServista API enables developers to integrate our powerful news verification and analysis algorithms into any software platform.

Submit any URL or raw text and retrieve structured information and trustworthiness metrics, such as TrustLevel and Patient Zero. Discover article links and relationships, similar content and get access to statistics and social media performance metrics.



TrustLevel

TrustLevel is a metric that indicates the trustworthiness of an article, taking into account various factors: from the publisher type and credentials, to the presence of known named author, punctuation, semantic analysis of the content and link analysis with other articles.  The TrustLevel has a value range between 0% and 100%, with 0% meaning “not at all trustworthy article” and 100% meaning “fully trustworthy article”.

There are only two situations when an article receives a 0% TrustLevel:

  • If the article is published by a known satire website.
  • If the TrustServista processing pipeline was unable to extract content from the website and no semantic analysis could be performed.

A sub-metric of the TrustLevel is the Clickbait score that indicates the probability of article to be “clickbait”. The Clickbait score has values between 0% – 100%, with 100% meaning “there is a 100% chance of this article to be clickbait”.

trustscore

An Example of an Article Graph

article graph