Bi-NAS Towards Effective and Personalized Explanation for Recommender Systems via Bi-Level Neural Architecture Search

Jul 16, 2026

Recommender systems must provide not only accurate suggestions but also effective, personalized explanations that strengthen user engagement, trust, and decision quality. The paper proposes Bi-level Neural Architecture Search (Bi-NAS), which optimizes explanation components by exploring both intra-layer and inter-layer design spaces. Bi-NAS refines cross-attention and feature interaction functions and augments generation with large language models using zero-shot prompting. Experiments on four real-world datasets show improved recommendation accuracy and substantially better explanation effectiveness, with data and code released publicly.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

Introduction

Motivation for Effective Explanations

Explainable Recommendation Approaches

Bi-NAS: Towards Effective and Personalized Explanation for Recommender Systems via Bi-Level Neural Architecture Search arXiv :2607 .0 1387v 1 [ cs .IR] 1 Jul 2026 Longfeng Wu Virginia Tech Blacksburg, VA, USA[longfengwu@vt.edu](longfengwu@vt.edu) Bhanu Pratap Singh Rawat Amazon Sunnyvale, CA, USA [rawabhan@amazon.com](rawabhan@amazon.com) Yao Zhou Google Mountain View, CA, USA [yaozhoucosmos@google.com](yaozhoucosmos@google.com) Lecheng Zheng Virginia Tech Blacksburg, VA, USA [lecheng@vt.edu](lecheng@vt.edu) Tong Zeng Virginia Tech Blacksburg, VA, USA [tongzeng@vt.edu](tongzeng@vt.edu) Giovanni Seni Amazon Sunnyvale, CA, USA [gseni@amazon.com](gseni@amazon.com) Zhimin Peng Amazon Sunnyvale, CA, USA [zmpeng@amazon.com](zmpeng@amazon.com) Dawei Zhou Virginia Tech Blacksburg, VA, USA [zhoud@vt.edu](zhoud@vt.edu) Abstract—Recommender systems are vital in helping users navigate vast amounts of information, offering personalized suggestions and effective explanations for these recommendations. While previous efforts have attempted to provide such explanations, evaluating their effectiveness across various scenarios remains a challenge. Enhancing these explanations is essential for improving user engagement, trust, and decision-making. To facilitate effective explanations within the recommender system, we propose a Bi-level Neural Architecture Search (Bi-NAS) framework to optimize explanations. This approach simultaneously refines cross-attention mechanisms and feature interaction functions by exploring both intra-layer and inter-layer design spaces. Furthermore, we integrate Large Language Models (LLMs) to enhance explanation generation, leveraging zero-shot prompting to produce more effective and personalized justifications. By aligning user feature preferences with item quality scores, our approach ensures that explanations reflect both user intent and item attributes, improving transparency and reasoning depth. Extensive evaluations on four real-world datasets demonstrate that Bi-NAS not only boosts recommendation accuracy but also significantly improves the effectiveness of explanations for recommender systems, providing users with clear and reliable insights into the suggestions they receive. Meanwhile, we publish our data and code at [https://github.com/wulongfeng/Bi-NAS.git](https://github.com/wulongfeng/Bi-NAS.git). Index Terms—Explainable Recommendation, Neural Architecture Search I. INTRODUCTION In the digital age, with the explosion of information and data, recommender systems have emerged as an essential tool for dealing with information overload [1] . These systems are designed to provide users with personalized and relevant content, which can help cater to their preferences and needs, stream their decision-making process, and improve service efficiency [2] . However, users may not be satisfied with the results they receive without clear and meaningful explanations, as shown in Figure 1 . Providing effective explanations for the recommendation results can significantly enhance the users’experience, foster their engagement and trust, and boost their loyalty and stickiness to the service and products. Meanwhile, an effective explanation could also be useful for the users to improve their decision-making accuracy and efficiency [3] or increase users’ satisfaction with the explanations [4] . Fig. 1. Illustration of recommender system with effective explanation. The reviews on the left are written by one user with their key aspects and opinions highlighted in blue. Similarly, each item will also receive multiple reviews from various users, with its key aspects and opinions marked in pink or blue. The word clouds are generated based on the user’s perspectives, with each word’s size in the word cloud reflecting its frequency, as the user mentioned. Early explainable methods, such as collaborative filtering (CF), generated simple justifications like “users who liked X also liked Y” [5, 6] .

FAQs

What evidence supports the effectiveness of Bi-NAS?

Extensive evaluations on four real-world datasets show that Bi-NAS improves recommendation accuracy and significantly enhances the effectiveness of explanations.

Jul 16, 2026

Upload by Lucas Martin