Item
GenUIne CRS: A Conversational Recommender with RAG
- Title
- GenUIne CRS: A Conversational Recommender with RAG
- Description
- In this demo, we show an adaptable proof-of-concept web application for a conversational recommender system which supports domain-specific user interface (UI) elements. The application uses a Large Language Model (LLM) for preference elicitation, content-based recommendation, and justification of the outcomes in different explanation styles. Through this interactive example, we aim to highlight three often overlooked aspects in current research on conversational recommender systems: (i) chatbots can go beyond purely text-based interaction patterns by implementing custom UI elements; (ii) function calling is a powerful feature to develop flexible, semi-structured interfaces with LLMs; (iii) due to their general knowledge and RAG extensibility, LLMs can be employed as convincing content-based recommenders to rapidly develop and test interfaces and explanation styles.
- Creator
- Ulysse Maes
- Contributor
- Communication Sciences | Faculty of Economic and Social Sciences and Solvay Business School
- Date
- 14 Oct 2024
- Type
- Software
- Language
- en
- Source
- https://researchportal.vub.be/en/publications/genuine-crs-a-conversational-recommender-with-rag/
- Rights
- Rights not stated in the record; consult the VUB Research Portal / rights holder
- Educational Level
- Higher education | research
- Interactivity Type
- Active
- Item sets
- Vrije Universiteit Brussel