Improving the Understanding of Arabic Political Text Through Emotion Analysis and Text Mining
Amjad Nassar, Mira Tzoreff, Tsvi Kuflik, Ilai Alon · Journal on Computing and Cultural Heritage · 2026
In this study, we explored the possibility of using text mining techniques to identify the emotive component of political articles written in Arabic. We use word embedding algorithms and a precompiled emotive lexicon to represent political terms and documents on a classical spectrum of emotions (anger, disgust, fear, joy, surprise, and sadness). Our hypothesis is that this methodology will enable us to identify noticeable emotive differences in the political lexicon between different political groups and periods, thereby contributing to improved understanding of historical and political phenomena (Giachanou et al., 2019).We demonstrate our methodology using a specific case study – examining articles related to the Muslim Brotherhood during the Arab Spring (2011–2012) and the period immediately followed (2013–2020). We analyze the emotive elements in Egyptian political vocabulary as employed in both anti- and pro-Muslim Brotherhood newspapers. Our results show that overt and covert emotions in the pro-Muslim Brotherhood newspapers changed from positive to negative in tandem with the political change from the period when the Muslim Brotherhood formed the government (Mohamed Morsi’s presidency) to that of President Abdel Fattah El-Sisi. In contrast, the emotions expressed in the opposition newspapers changed from mostly negative to mostly positive between the same two periods.Based on our case study, we suggest insofar as our method can provide consistent and reliable results for evaluating emotions related to trust and political terms in printed politics related media, it may also be applied in additional domains. That said, the results are still somewhat speculatory and should be explored in further detail in future work.