Since the beginning of the year media industry is under intense pressure finding out ways to verify news content and get way from propaganda. Even social media giants like Facebook and Google have come out in open to help media organizations collaborate this epidemic that is slowly crippling the media space. While there are only a few steps that have been inching towards a solution, one such is by a Romanian technology company Zetta Cloud that has developed TrustServista, a newly launched unique digital investigation tool that automatically determines the origin and trustworthiness of online news stories – thanks to the Google DNI grant.
Though the tool though is still in its early phase, however, the solution it offers and the interest it has gained across media space is attention grabbing. Even BuzzFeed’s Craig Silverman who has been studying fake news for years recently noted that “it’s one to watch.” So we thought to learn more about the tool and spoke with Zetta Cloud’s co-founder George Bara to know TrustServista’s future plan. Below are the excerpts from the interview.
Q: With what kind of technology is TrustServista solving the problems of fake-news and performing news-verification?
George Bara (GB): As opposed to most recent approaches for fighting fake news focused on crowdsourcing or even cross-newsroom initiatives, TrustServista proposes a fully automated news verification process, relying on text analytics, natural language processing and machine learning algorithms. One of the main tasks performed by the software is to determine the origin of information referenced by news articles (which we call “patient zero”), using link analysis concepts implemented using complex content categorization, entity extraction and similarity detection algorithms that can currently handle any English language news article published on the web. It then relies on automatically extracting information from the content, such as sentiment, objectivity and context setting, in order to determine its trustworthiness. Every operation that TrustServista performs – from data collection to data processing, is done automatically with the help of in-house of 3rd party text processing software modules.
Q: When can we expect a beta release?
GB: The beta release can be expected in less than a month. Since the “alpha” launch, which was at the beginning of February 2017, we have collected a lot of user feedback that helped us fix some of the initial issues and improve our product. We are currently working on increasing the platform’s data processing performance in order to to able to process as many websites as possible. In the next months after the beta we will release TrustServista 1.0, a production version that will offer SLAs and even an API for integration purposes, and could be purchased by media organizations, academia and other agencies.
Q: How can newsrooms benefit from TrustServista and the kind of business model you have?
GB: TrustServista’s unique selling proposition is to significantly reduce investigation times in order to produce high quality trusted content. Even established news agencies can make mistakes due to the high volumes of information produced with high velocity every day and the shortage of resources. TrustServista automatically determines the origin of information and scores each referenced news article based on its trustworthiness, so that media professionals can identify very quickly where the information is coming from and whether the sources are trustworthy or not. The solutions is available as Software as a Service (SaaS), easy to adopt with little to none involvement from IT. TrustServista will offer different subscriptions, depending on the number of users and features that are required, a model that can scale easily as customers and their needs evolve.
Q: The world is soon about to witness the French and German elections, tools like TrustServista could be in high demand. How can TrustServista help perform fact-checking and prevent propaganda?
GB: Elections are very susceptible to fake news, as we have witnessed in last year’s US presidential elections. It is also a period when this phenomena becomes extremely visible, resulting in a gradual but steady loss of the public’s trust in media organizations and even influencing election results. TrustServista is designed in such a way that is it can be used not only by media organizations and content creators, but also by content distributors (such as social media platforms) and readers. One of the most important uses for TrustServista would be to automatically find and flag all news articles that reference untrusted or false information and prevent their distribution and “viralization” across social media, thus limiting their influence and reach.
Q: What are your future plans with TrustServista?
GB: Looking into the future – we definitely plan to offer support for non-English content, such as Spanish, German, French, Arabic and Russian, in order to have a global industry impact. This expansion into foreign languages will be combined with improving our trustworthiness score algorithm, using advanced natural language processing algorithms and machine learning techniques that will be able to handle most click bait scenarios and even complex propaganda. Eventually the platform will allow user collaboration for investigations performed across multi-journalist teams, but most of the data collection and processing will still be performed automatically. Our main goal is to offer a unique solution that will automate most – if not all – of the tasks performed by fact checkers and investigators supporting newsrooms in creating trusted content.