Unveiling Persian Market Dynamics: A Comprehensive Analysis of Consumer Demand Using NLP Techniques with Explainable Artificial Intelligence

The Persian language, spoken by more than 110 million people globally, ranks as the 18th most spoken language worldwide.However, the development of Natural Language Processing (NLP) techniques for Persian faces challenges due to limited annotated datasets, dialect diversity, and its complex morphology and syntax.This study addresses this gap by using NLP techniques to analyze market demand in Persian.The research includes sentiment analysis, named entity recognition (NER) and gender prediction to offer insights into consumer preferences.The paper proposes ParsBERT model for sentimental analysis of Persian texts.The data set is collected from various social media platforms, focusing on user reviews of laptop brands and labeled with negative, positive and neutral comments.The ParsBERT model and deep learning methods are analyzed using precision, recall, f1 score, and precision metrics.The results show that the ParsBERT model achieved an accuracy of 98%, which is higher than other machine learning and deep learning models for determining the sentiment of social media comments.Additionally, two Explainable AI (XAI) methods, Local Interpretable Model-Agnostic Explanations (LIME) and Shapley Additive Explanations (SHAP), are used to clarify the decision-making process of the proposed model.The results show most preferred brand among both male and female consumers.This research provides valuable information for businesses and policymakers seeking to understand and meet the demands of the Farsi-speaking market.