# Design aesthetics recommender system based on customer profile and wanted affect

**Authors:** Brahim Benaissa, Masakazu Kobayashi, Keita Kinoshita  
**arXiv:** [2301.10984](https://arxiv.org/abs/2301.10984)  
**Date:** 2023-01-26  
**Summary page:** https://www.summarizepaper.com/en/paper/2301.10984v1/

## Key points

- Profiling customers based on preferred product design and desired effects
- Importance of product recommendation systems in online commerce
- Enhancement of recommendation systems with big data and advanced deep learning techniques
- Current limitation of choice definition in consumer profiling
- Focus on profiling customers based on preferences for product design and desired effects
- Examination of individual Kansei (aesthetic) aspects of vase designs and their influence on consumer responses
- Literature review to inform selection of personal aspects considered in the study
- Development of a representative consumer model using deep learning techniques as the core of the recommendation system
- Proposed system prompts new consumers to specify desired effect using Kansei adjectives
- System recommends aesthetic design most likely to evoke the specified effect
- Contribution to expanding existing product recommendation systems by incorporating customer profiles based on preferred product design and desired effects
- Aim to enhance user satisfaction by recommending aesthetic designs that align with individual preferences and desired effects.

## Summary

The paper titled "Design aesthetics recommender system based on customer profile and wanted affect" explores the potential of profiling customers based on their preferred product design and desired effects. The authors highlight the importance of product recommendation systems in online commerce, which have been further enhanced by big data and advanced deep learning techniques. These systems rely on consumer profiling to predict consumer interests based on their past choices and the choices of similar consumers. However, the current definition of choice is limited to quantifiable data such as product features, cost, and type. To address this limitation, the study focuses on profiling customers based on their preferences for product design and desired effects. Specifically, the authors examine individual Kansei (aesthetic) aspects of vase designs to understand how they influence consumer responses. To establish a comprehensive understanding of consumer preferences, the researchers conduct a literature review that informs their selection of personal aspects considered in the study. They then develop a representative consumer model using deep learning techniques as the core of their recommendation system. The proposed system prompts new consumers to specify the desired effect they are looking for using Kansei adjectives. Based on this input, the system recommends an aesthetic design that is most likely to evoke that particular effect. In conclusion, this paper contributes to expanding existing product recommendation systems by incorporating customer profiles based on preferred product design and desired affects. By leveraging deep learning methods and Kansei adjectives, the proposed system aims to enhance user satisfaction by recommending aesthetic designs that align with individual preferences and desired effects.