LLM_RS

This repository explores the integration of Large Language Models (LLMs) into recommendation systems (RS) and related applications. It is organized into three primary modules: LLM for Embedding, LLM as RS, and LLM Message Generator. Each module serves a distinct role in leveraging LLMs for personalized content recommendations and customer interaction.

English Descriptions

1. LLM for Embedding

This module explores how content items can be represented as embeddings using LLMs and applied to the SASRec recommendation system. The SASRec algorithm is a sequential recommendation system that leverages self-attention mechanisms to predict user preferences based on their interaction history.

2. LLM as RS

This module demonstrates how LLMs can act as recommendation agents by suggesting the next content item for users. Instead of solely relying on traditional algorithms, LLMs are used as intelligent agents capable of reasoning and making recommendations.

3. LLM Message Generator

This module focuses on generating persuasive recommendation messages. After the LLM-based recommendation system identifies suitable content, this module crafts convincing and personalized messages to explain the recommendation reasons to customers.

Korean Descriptions

LLM을 RS에 다양한 방법으로 적용하는 프로젝트입니다.

• LLM for Embedding: 각 콘텐츠 아이템을 임베딩을 LLM으로 진행한 후 기존 SASRec 추천 시스템에 적용하는 방법
  • SASREC 알고리즘 출처: https://github.com/yehjin-shin/BSARec
• LLM as RS: LLM Agent에게 다음 콘텐츠를 추천하는 추천 시스템 역할을 수행하게 하는 방법
• LLM Message Generator: LLM Agent에게 추천 시스템이 추천한 콘텐츠의 추천 이유에 대해서 고객들에게 설득할만한 추천 문구를 만드는 역할을 수행하게 하는 방법

You can feel free to contact @uyunho99 for more information.