Predicting the Political Affiliation of Moroccan Parties Through Analysis of Electoral Programs

Our study originates from the lack of scientific analysis in political life. Recent advancements in Natural Language Processing enable a comprehensive examination of political discourse and party programs. The initial step in a voter's decision-making process is distinguishing the political orientation of parties, emphasizing the need for clear distinctions in the programs presented by these political entities. We conducted an analysis of the political programs of the six primary political parties in Morocco, situated in North Africa. Our objective was to determine whether their political proposals could be effectively classified within the progressive–conservative framework. Two distinct methods were employed for this assessment. The first method involved calculating the frequency of terms and their cosine similarity. The properties of word embeddings facilitate the representation of similar words, thereby allowing for the grouping of related topics. The second method utilized prompt engineering with the Language Model “LLaMA”. The findings indicate that Moroccan political parties can be effectively classified into the progressive–conservative framework based on an analysis of their respective programs.