A new scientific paper has been published that takes on the challenge of assessing the neutrality of Wikipedia articles about companies, using advanced sentiment analysis tools. This publication represents an important step toward monitoring the neutrality of one of the most influential sources of knowledge on the Internet and may be useful to anyone interested in research on information quality.
One of the key quality criteria for Wikipedia articles is adherence to the Neutral Point of View (NPOV) principle, which requires that content be free from subjective opinions and bias. Yet, the collaborative nature of editing means that unintentional bias can creep into articles, especially those about companies, which attract attention from fans, critics, and company representatives alike. Detecting such bias is no easy feat. Standard sentiment analysis tools are often designed for short, opinionated texts, like tweets or reviews – not for the carefully crafted, sometimes dry prose of Wikipedia. Applying these tools to encyclopedia articles requires not just technical skill, but an understanding of both language and the editorial dynamics of Wikipedia.
The importance of this work goes far beyond academic curiosity. Wikipedia is frequently used by journalists, investors, and everyday consumers researching companies. The presence of bias (whether positive or negative) can shape public perception, influence decisions, and even affect markets. That’s why it’s crucial to develop reliable methods for assessing the neutrality of Wikipedia content.
In this new study, researchers compared both traditional lexicon-based sentiment analysis tools (like TextBlob and VADER) and state-of-the-art transformer-based models (including RoBERTa, DistilBERT, and PySentimiento). They rigorously selected, cleaned, and analyzed Wikipedia articles about companies, then aggregated sentiment scores using three different methods. The study not only compared overall sentiment but also broke down results by industry and Wikipedia article quality class.
The publication “Sentiment Analysis of Wikipedia Articles About Companies: A Comparison of Different Models” was published on the Springer website. Authors of the publication: Dr. Milena Stróżyna, Dr. Włodzimierz Lewoniewski, Izabela Czumałowska, Aleksandra Wojewoda.
Sources: kie.ue.poznan.pl, ue.poznan.pl