[4F3-GS-10n] AI応用：建設・不動産

[4F3-GS-10n-04] Graph-Based Analysis of a Large-scale Attractiveness Dataset for Real Estate Floor Plans

〇Taro Narahara¹, Xueting Wang², Toshihiko Yamasaki² (1. New Jersey Institute of Technology, 2. The University of Tokyo)

Keywords:Real Estate Floor Plans, Attractiveness Computing, Crowdsourcing

This paper presents the analysis of the large-scale dataset based on the attractiveness of floor plans using nine statements that explore qualitative and functional values of apartments, such as a level of comfort, modernity, etc., using crowdsourcing. Understanding relationships between real estate floor plans and their inherent qualities in a quantitative data format would be important to obtain deep neural networks that could extract features related to the attractiveness of apartments. We developed an effective method to extract emergent subgraphs that contributed to higher or lower scores for such criteria. Results revealed several characteristics that matched with our general preferences for apartments.