LLM RFM and Partnership Rule Live Stream Interaction Suggestion System

Abstract

It is an interaction suggestion system developed for live streaming services and subscription-based content platforms in the telecommunications sector, and its features include: a data collection module (1) that collects users' live streaming viewing times, a user database (2) that stores messages sent by users, expenses incurred, and streaming viewing times, a user database network server (3) which is the server hardware on which the database application is installed, a user behavior analysis interface with RFM model that analyzes user behavior, expenses incurred, and last interaction times (4), an RFM model user behavior analysis interface network server (5) which is the server hardware on which the RFM model user behavior analysis interface application runs, and a community rules module (6) that analyzes message, comment, emoji, gift sending, and click behaviors collected from past broadcasts.The Live Broadcast Instant Interaction Prediction Module includes the following: the association rules module network server (7), which is the server hardware running the association rules module software; the Live Broadcast Instant Interaction Prediction Module (8), which predicts the probability of instant interaction in the live broadcast using the RFM results and features generated from association rules; the Live Broadcast Instant Interaction Prediction Module network server (9), which is the server hardware running the Live Broadcast Instant Interaction Prediction Module software; the LLM-based content and interaction suggestion module (10), which generates real-time, natural language automated suggestions using the prediction results and association rules; and the LLM-based content and interaction suggestion module network server (11), which is the server hardware running the LLM-based content and interaction suggestion module software.