FUNIBER’s Director of Admissions presents an approach to categorize YouTube videos

FUNIBER’s Director of Admissions presents an approach to categorize YouTube videos

Dr. Monica Gracia, International Director of Admissions at the Fundación Universitaria Iberoamericana (Iberoamerican University Foundation, FUNIBER), participates in a study that develops an approach based on artificial intelligence to categorize YouTube videos, in order to facilitate the user’s search for information.

In a world where 300 hours of video are uploaded to YouTube every minute, finding relevant content can be like looking for a needle in a digital haystack. But what if artificial intelligence could transform this experience, ranking videos with near-perfect accuracy?

Over the past decade, video content on web platforms has grown significantly, driven by free access to sites like Facebook and YouTube. Founded in 2005, YouTube has quickly become the second largest video sharing platform, with around 37 million channels. This platform allows users to watch, share, comment and upload videos, making it a constant source of information on a variety of topics. Every minute, a huge number of videos are uploaded to YouTube, covering a wide range of topics. These videos are easily shareable through social networks, websites and email, and can be integrated into other web pages. Proper categorization of videos is crucial to improve their visibility and facilitate their searchability, using tags and categories to optimize their classification.

Deep learning models have been instrumental in image processing and video classification. These techniques have proven to be effective in natural language processing (NLP) tasks, such as semantic analysis and sentence modeling. Convolutional neural networks (CNNs), originally developed for computer vision, have shown success in PLN. In video classification, both visual and textual features can be used, with text-based categorization being less computationally complex, although limited when textual information is sparse.

Despite these challenges, text-based video categorization offers advantages in terms of computational simplicity. This process, known as text classification, organizes documents into specific groups, with binary classification being the most common. The objective is to characterize a document in order to assign it to the appropriate category, without extracting additional information. In this sense, the study presents an innovative artificial intelligence-based approach to address this challenge, which consists of a deep convolutional neural network (DCNN) focused on using textual features such as titles, descriptions and user tags to categorize videos.

The results showed an outstanding ability to classify YouTube videos with an accuracy of 96% and an area under the curve (AUC) of 99% on receiver operating features. This significantly outperforms existing methods, offering a powerful tool to suggest relevant videos to users and organize them by category more effectively.

In addition to DCNN, the study also evaluated the performance of recurrent neural networks (RNN) and closed recurrent units (GRU), as well as logistic regression models, support vector machines, decision trees, and random forests. However, DCNN emerged as the most effective in the task of video categorization.

This breakthrough not only promises to improve the user experience on video platforms, but could also have significant implications for digital content management in general. By enabling more accurate and efficient categorization, AI-based tools like DCNN could transform how online content is interacted with, making it easier to access relevant information in a sea of data. In the near future, searching for videos on YouTube could be just as simple and accurate.

If you want to learn more about this study, click here.

To read more research, consult the UNEATLANTICO repository.

The Iberoamerican University Foundation (FUNIBER) promotes several master’s degree programs in the area of technology, such as the Master’s Degree in Strategic Management in Software Engineering. This master’s degree prepares you to be an agent of transformation within organizations. You will learn to integrate software engineering into business strategies, developing essential competencies in strategic direction, leadership, and technology management. Don’t miss the opportunity to become a strategic leader in software engineering! Enroll now and take your career to the next level.