Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Retrieval-Augmented Conversational Recommendation with Prompt-based Semi-Structured Natural Language State Tracking

Sara Kemper, Justin Cui, Kai Dicarlantonio, Kathy Lin, Danjie Tang, Anton Korikov, Scott Sanner

TL;DR

The paper tackles the challenge of mapping diverse natural-language user preferences to relevant items in conversational recommendation when metadata is incomplete. It presents RA-Rec, a Retrieval-Augmented, LLM-driven dialogue state tracking framework that uses a prompt-driven loop to classify intents and maintain a semi-structured NL dialogue state. Leveraging late fusion Reviewed Item Retrieval over item reviews and metadata, RA-Rec generates personalized recommendations, explanations, and QA with a scalable pipeline that employs a TAS-B encoder and FAISS for efficient retrieval. Demonstrated in the restaurant domain with Yelp data and open-source artifacts, RA-Rec shows practical feasibility and provides a path toward domain adaptation, active elicitation, and multi-aspect reasoning in ConvRec systems.

Abstract

Conversational recommendation (ConvRec) systems must understand rich and diverse natural language (NL) expressions of user preferences and intents, often communicated in an indirect manner (e.g., "I'm watching my weight"). Such complex utterances make retrieving relevant items challenging, especially if only using often incomplete or out-of-date metadata. Fortunately, many domains feature rich item reviews that cover standard metadata categories and offer complex opinions that might match a user's interests (e.g., "classy joint for a date"). However, only recently have large language models (LLMs) let us unlock the commonsense connections between user preference utterances and complex language in user-generated reviews. Further, LLMs enable novel paradigms for semi-structured dialogue state tracking, complex intent and preference understanding, and generating recommendations, explanations, and question answers. We thus introduce a novel technology RA-Rec, a Retrieval-Augmented, LLM-driven dialogue state tracking system for ConvRec, showcased with a video, open source GitHub repository, and interactive Google Colab notebook.

Retrieval-Augmented Conversational Recommendation with Prompt-based Semi-Structured Natural Language State Tracking

TL;DR

The paper tackles the challenge of mapping diverse natural-language user preferences to relevant items in conversational recommendation when metadata is incomplete. It presents RA-Rec, a Retrieval-Augmented, LLM-driven dialogue state tracking framework that uses a prompt-driven loop to classify intents and maintain a semi-structured NL dialogue state. Leveraging late fusion Reviewed Item Retrieval over item reviews and metadata, RA-Rec generates personalized recommendations, explanations, and QA with a scalable pipeline that employs a TAS-B encoder and FAISS for efficient retrieval. Demonstrated in the restaurant domain with Yelp data and open-source artifacts, RA-Rec shows practical feasibility and provides a path toward domain adaptation, active elicitation, and multi-aspect reasoning in ConvRec systems.

Abstract

Conversational recommendation (ConvRec) systems must understand rich and diverse natural language (NL) expressions of user preferences and intents, often communicated in an indirect manner (e.g., "I'm watching my weight"). Such complex utterances make retrieving relevant items challenging, especially if only using often incomplete or out-of-date metadata. Fortunately, many domains feature rich item reviews that cover standard metadata categories and offer complex opinions that might match a user's interests (e.g., "classy joint for a date"). However, only recently have large language models (LLMs) let us unlock the commonsense connections between user preference utterances and complex language in user-generated reviews. Further, LLMs enable novel paradigms for semi-structured dialogue state tracking, complex intent and preference understanding, and generating recommendations, explanations, and question answers. We thus introduce a novel technology RA-Rec, a Retrieval-Augmented, LLM-driven dialogue state tracking system for ConvRec, showcased with a video, open source GitHub repository, and interactive Google Colab notebook.
Paper Structure (14 sections, 2 figures, 3 tables)

This paper contains 14 sections, 2 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Background and Related Work
  3. Dialogue State Tracking
  4. Reviewed Item Retrieval
  5. Retrieval Augmented Conversational Recommendation
  6. Prompt-Driven Intent Classification
  7. Semi-Structured NL Dialogue State Tracking
  8. State Elements
  9. Action Selection
  10. Retrieval-Augmented Recommendation and Explanation
  11. Retrieval-Augmented Question Answering
  12. RA-Rec System Summary
  13. Demonstration Details
  14. Future Work

Figures (2)

  • Figure 1: The RA-Rec prompt-driven dialogue state tracking loop. LLM prompting is used for multi-label intent classification and for updating a JSON semi-structured NL state which tracks user preferences and other key dialogue elements. The state keys provide an easily configurable structure, while LLM-generated state values can capture nuanced NL expressions.
  • Figure 2: Retrieval-augmented recommendation and explanation using late fusion RIR. First, preferences in the dialogue state are used to generate a NL query. Then, query and review embeddings are scored using dot product similarity, and the top review scores for each item are averaged (fused) into an item score. Finally, the top items' most relevant reviews and metadata are used to generate a recommendation and explanation of how the item satisfies the preferences in the state.