کنفرانس بین المللی فناوری اطلاعات و دانش

صفحه اصلی / شانزدهمین کنفرانس بین المللی فناوری اطلاعات و دانش

AI-Powered Beauty Insights: Sentiment Analysis in a Low-Resource Language

نویسندگان :

Sajedeh Talebi¹ Neda Abdolvand² Fatemeh Mahdian³

1- دانشگاه الزهرا(س) 2- دانشگاه الزهرا(س) 3- دانشگاه الزهرا(س)

کلمات کلیدی :

sentiment analysis،low-resource language،cosmetic reviews،ensemble learning،Persian natural language processing،Digikala

چکیده :

This study uses artificial intelligence (AI) and natural language processing (NLP) to analyze 24,256 Persian-language cosmetic reviews from Digikala, Iran’s leading e-commerce platform, to uncover consumer preferences for skincare products, including Exfoliating Cream, Anti-Wrinkle Cream, Eye Moisturizer, and Sunscreen Cream. Using Pars-BERT, a transformer model tailored for Persian, and preprocessing tools like Hazm and FarsiYar, our pipeline employs ensemble learning and data augmentation (e.g., SMOTE, a technique to balance datasets) to identify “effectiveness” and “price” as key drivers of purchase decisions. Visualizations, including word clouds and UMAP (a dimensionality reduction technique), enable brands to optimize product positioning and personalize marketing campaigns in low-resource language markets. This approach advances Persian NLP and provides e-commerce businesses with data-driven tools to enhance sales and customer satisfaction.