Hybrid Recommendation System based on Natural Language Processing

In recent times, the online retail industry has seen active use of natural language processing (NLP)-based product analysis and recommendation systems that utilize seller-provided product information and customer review data.This study developed an NLP-based recommendation system by combining the data characteristics of a specific socks product, with customer review data.The system analyzes and categorizes seller-provided product information for socks (e.g., gender, length, pattern, color, price) and previous customer reviews (e.g., quality, delivery) to recommend the most suitable products.The seller's product information was processed through text-based learning, while the review data was processed using sequence-based learning, enhancing the accuracy of the recommendations.This study introduces a hybrid approach that employs the TF-IDF model for keyword association analysis and the GRU model for sentiment analysis.The sock recommendation system was developed using review data sourced from the