LERA: LLM-Enhanced RAG for Ad Auction in Generative Chatbots
Researchers have developed LERA, a new framework for integrating advertising auctions into large language model chatbots. Unlike previous methods that relied solely on text similarity for ad selection, LERA uses a two-stage process. First, it employs embedding-based filtering to narrow down potential advertisers, and then it leverages the LLM itself to generate refined relevance scores. This approach aims to improve ad accuracy and diversity while managing latency. AI
IMPACT This framework could enable more effective monetization of LLM chatbots by improving ad relevance and diversity.