Dr. Mónica Gracia, international director of Admissions at the Iberoamerican University Foundation (FUNIBER), and Dr. Eduardo Silva, executive director of the same Foundation at its headquarters in Guatemala, are conducting a joint study with other professionals to improve the automatic detection of fake news through machine learning. This work demonstrates the effectiveness of a hybrid and multi-view approach to analyzing news content in digital environments.
In recent years, misinformation has grown rapidly on social media and in digital media. Messages that mix truth with falsehoods influence public discourse, erode trust, and complicate decision-making, especially during election periods or health crises such as COVID-19. Therefore, having automatic tools that help identify and curb the spread of misleading content has become a priority for newsrooms, digital platforms, and authorities.
Until now, many solutions have been based on examining only the text or using very complex models. Although they have been a step forward, they often have difficulty adapting to new contexts, handling controversial texts, or explaining why they make a decision. They also tend to «flatten» all the information into a single block, losing important nuances of language and subject matter.
The study proposes a different and easy-to-understand approach: analyzing each news item from three complementary angles and then combining these «opinions» in an intelligent way. First, the text itself is observed (frequent words and expressions), then how it is written (comprehensibility, emotional tone, use of proper names, and grammatical structure), and finally, what it is really about (dominant themes and general meaning). A specialized model is trained for each perspective; in the end, an “arbitrator” brings the three together and makes a more accurate decision than any of them could make separately.
To test it, the team worked with tens of thousands of articles already classified as real or fake and used a rigorous evaluation that repeats the training and testing in ten different rounds to avoid chance. In addition, they checked whether the system remains stable when the texts undergo minor alterations (for example, deleting or changing the order of some words) and whether it can transfer what it has learned to a different set, consisting of short political phrases.
The results are particularly robust. In the main set, the system is correct 99.94% of the time and outperforms both models that look at a single perspective and others that mix everything in a single step. It also outperforms a very powerful reference based on deep neural networks. When evaluated with the set of short sentences, it maintains a very high level of accuracy (around 97%), indicating that it generalizes well even when the type of text changes. In endurance tests, accuracy remains above 97% even if some of the words are deleted, swapped, or repeated, a sign that if one of the views loses information, the other two compensate for it.
This proposal strikes an appropriate balance between effectiveness and computational cost: it improves key metrics without requiring heavy infrastructure, making it easy to adopt in resource-constrained environments.
If you would like to learn more about this study, click here.
To read more research, check out the UNEATLANTICO repository.
The Iberoamerican University Foundation (FUNIBER) offers scholarships to study the Master’s Degree in Strategic Management in Information Technology. This program prepares you to lead digital transformation by aligning technology and business: systems planning and governance, project and service management, cybersecurity, and data for decision-making. Take the next step in your life and study this master’s degree to lead the technological future of business.