Machine Learning Deciphers News Language to Predict Country’s Peace Level

Reading Time: 2 minutes

A recent study in the open-access journal PLOS ONE, conducted by Larry Liebovitch and Peter T. Coleman of Columbia University, along with their colleagues, introduces a “peace index” determined by analyzing word frequencies in mainstream news media. This index measures the level of peace in a given country. The language used in media not only mirrors a culture’s perspective but also influences its inhabitants. While the impact of “hate speech” on inciting violence is widely recognized, this study delves into the concept of “peace speech” and its role in cultivating and maintaining peace.

In their research, Liebovitch and his team employed five established peace indices to assess 18 countries categorized as high-peace, intermediate-peace, or low-peace. They examined 723,574 English-language media articles from these nations, all sourced locally. By applying a machine learning model to media content from high- and low-peace countries, they identified words associated with different levels of peace.

Overall, low-peace countries exhibited a higher prevalence of words related to government, order, control, and fear (e.g., government, state, law, security, court). In contrast, high-peace countries featured more words related to optimism and enjoyment (e.g., time, like, home, believe, game). When the machine learning model was applied to media from the intermediate-peace countries, it accurately identified their intermediate peace levels.

The authors acknowledge limitations in their data, primarily that it was derived from English-language sources, impacting its reliability in evaluating countries where English is less commonly used for news communication. They also recognize potential bias within the existing peace indices. Nonetheless, this study lays the groundwork for further exploration of linguistic differences between high-peace and low-peace cultures. The authors conclude that machine learning can effectively identify words in local news media that reflect a country’s peace level, with the focus shifting from government control in less peaceful countries to personal preferences and daily activities in more peaceful nations. High-peace countries also exhibit a greater diversity of terms compared to low-peace countries.

 

Source NeuroScienceNews

Neurologica
Author: Neurologica