Open versus Closed: A Comparative Empirical Assessment of Automated News Article Tagging Strategies
Michał Pogoda, Marcin Oleksy, Konrad Wojtasik, Tomasz Walkowiak, Bartosz Bojanowski · Procedia Computer Science · 2023
Automatically tagging news articles is a fundamental task for indexing and analyzing news streams. Most news portals a form of article tagging ontology, which is a predefined set of tags or keywords that are assigned to news articles to improve their discoverability and relevance. However, manually categorizing and tagging news articles using an ontology is a time-consuming and labor-intensive task, especially for news portals that publish a large volume of articles each day. The automation of the news article tagging process using machine learning techniques has emerged as a promising solution to address the challenge of manually categorizing and tagging a large volume of articles each day. In this paper, we explore the effectiveness of different machine learning approaches for automatically tagging news articles. Specifically, we train and compare several models using both closed-ontology and open-ontology approaches. We evaluate the performance of these models based on their ability to accurately categorize and tag news articles using comparative human evaluation. Our findings provide insights into the advantages and limitations of each approach and highlight the potential applications of automatic news article tagging.