Title: InteraRec: Interactive Recommendations Using Multimodal Large Language Models

Abstract: Weblogs, comprised of records detailing user activities on any website, offer valuable insights into user preferences, behavior, and interests. Numerous recommendation algorithms, employing strategies such as collaborative filtering, content-based filtering, and hybrid methods, leverage the data mined through these weblogs to provide personalized recommendations to users. Despite the abundance of information available in these weblogs, identifying and extracting pertinent information and key features necessitates extensive engineering endeavors. The intricate nature of the data also poses a challenge for interpretation, especially for non-experts. In this study, we introduce a sophisticated and interactive recommendation framework denoted as InteraRec, which diverges from conventional approaches that exclusively depend on weblogs for recommendation generation. This framework captures high-frequency screenshots of web pages as users navigate through a website. Leveraging state-of-the-art multimodal large language models (MLLMs), it extracts valuable insights into user preferences from these screenshots by generating a user behavioral summary based on predefined keywords. Subsequently, this summary is utilized as input to an LLM-integrated optimization setup to generate tailored recommendations. Through our experiments, we demonstrate the effectiveness of InteraRec in providing users with valuable and personalized offerings.