Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Diffusion-EXR: Controllable Review Generation for Explainable Recommendation via Diffusion Models

Ling Li, Shaohua Li, Winda Marantika, Alex C. Kot, Huijing Zhan

TL;DR

Diffusion-EXR tackles explainable recommendations by generating topic-focused reviews conditioned on pseudo user/item profiles and diffusion-based text embeddings. It integrates a DDPM framework within a lightweight Transformer to jointly predict ratings and produce natural explanations, while enabling user-guided keyword control. Empirically, it achieves state-of-the-art performance on explainability metrics and text quality on Amazon-CSJ and TripAdvisor, with competitive rating accuracy. This work demonstrates the practicality of diffusion models for transparent, justifiable recommendations and provides a reusable framework for controllable text generation in recommender systems.

Abstract

Denoising Diffusion Probabilistic Model (DDPM) has shown great competence in image and audio generation tasks. However, there exist few attempts to employ DDPM in the text generation, especially review generation under recommendation systems. Fueled by the predicted reviews explainability that justifies recommendations could assist users better understand the recommended items and increase the transparency of recommendation system, we propose a Diffusion Model-based Review Generation towards EXplainable Recommendation named Diffusion-EXR. Diffusion-EXR corrupts the sequence of review embeddings by incrementally introducing varied levels of Gaussian noise to the sequence of word embeddings and learns to reconstruct the original word representations in the reverse process. The nature of DDPM enables our lightweight Transformer backbone to perform excellently in the recommendation review generation task. Extensive experimental results have demonstrated that Diffusion-EXR can achieve state-of-the-art review generation for recommendation on two publicly available benchmark datasets.

Diffusion-EXR: Controllable Review Generation for Explainable Recommendation via Diffusion Models

TL;DR

Diffusion-EXR tackles explainable recommendations by generating topic-focused reviews conditioned on pseudo user/item profiles and diffusion-based text embeddings. It integrates a DDPM framework within a lightweight Transformer to jointly predict ratings and produce natural explanations, while enabling user-guided keyword control. Empirically, it achieves state-of-the-art performance on explainability metrics and text quality on Amazon-CSJ and TripAdvisor, with competitive rating accuracy. This work demonstrates the practicality of diffusion models for transparent, justifiable recommendations and provides a reusable framework for controllable text generation in recommender systems.

Abstract

Denoising Diffusion Probabilistic Model (DDPM) has shown great competence in image and audio generation tasks. However, there exist few attempts to employ DDPM in the text generation, especially review generation under recommendation systems. Fueled by the predicted reviews explainability that justifies recommendations could assist users better understand the recommended items and increase the transparency of recommendation system, we propose a Diffusion Model-based Review Generation towards EXplainable Recommendation named Diffusion-EXR. Diffusion-EXR corrupts the sequence of review embeddings by incrementally introducing varied levels of Gaussian noise to the sequence of word embeddings and learns to reconstruct the original word representations in the reverse process. The nature of DDPM enables our lightweight Transformer backbone to perform excellently in the recommendation review generation task. Extensive experimental results have demonstrated that Diffusion-EXR can achieve state-of-the-art review generation for recommendation on two publicly available benchmark datasets.
Paper Structure (19 sections, 13 equations, 1 figure, 5 tables)

This paper contains 19 sections, 13 equations, 1 figure, 5 tables.

Table of Contents

  1. Introduction
  2. Preliminary
  3. Problem Formulation
  4. Diffusion Model
  5. Methodology
  6. Self-attention Encoder
  7. Pseudo Persona and Profile Construction
  8. Self-Attention Mechanism
  9. Transformer-based Decoder
  10. Diffusion of Text Embeddings
  11. Rating Prediction
  12. Context Prediction
  13. Review Generation
  14. Multi-task Learning
  15. EXPERIMENTS
  16. ...and 4 more sections

Figures (1)

  • Figure 1: The architecture of Diffusion-EXR. Given user $u$ and item $i$, the goal of explanable recommendation is to generate a rating score $r_{(u,i)}$ and a justification in the form of a sentence $[w_1, w_2,\cdots,w_N]$.