LyFy: Enhancing Batik E-Commerce Live Streaming Through Real-Time Chat Filtering and Product Recommendation

Abstract

Live streaming has emerged as an essential tool for e-commerce, allowing sellers to engage with potential customers in realtime. However, the massive influx of comments during these sessions often includes a mix of useful product-related queries and irrelevant or distracting messages, which can overwhelm the presenter and reduce the effectiveness of the stream. In this paper, we propose LyFy, a browser-based extension designed to filter live chat messages and provide personalized product recommendations in real-time, specifically applied in Batik e-commerce to support the preservation and promotion of this unique cultural heritage of Indonesia. Our system uses a combination of natural language processing (NLP) and machine learning models to identify relevant comments, group similar queries, and offer product suggestions based on viewers' interests. We demonstrate the effectiveness of this system through a prototype implementation and evaluate its performance with qualitative feedback from streamers and users. The evaluation results indicate high user satisfaction, with over 51% of respondents rating LyFy as highly effective and 52% as highly efficient, making it a valuable tool for enhancing e-commerce live streaming interactions.