[1N5-GS-13] AI application: Machine learning and application (2)

[1N5-GS-13-03] Entity Resolution of Apartment Property Using Neural Networks

〇Youiti Kado¹, Takashi Hirokata¹, Koji Matsumura¹, Xueting Wang², Toshihiko Yamasaki² (1. At Home Lab Co., Ltd., 2. The Univ. of Tokyo)

Keywords:entity resolution, real estate, neural network, property information

Property information for apartment rooms must be linked to the correct apartment building to be used effectively. The work of aggregating property information belonging to the same building (entity resolution) is commonly executed by a rule-based process that statistically considers the similarity of attributes such as building name, number of floors, or year/month built. However, when property information is stored by room and registered by different businesses, the corresponding building information may be inconsistent, incomplete or inaccurate. Therefore, entity resolution using a rule-based method is insufficient and requires extensive manual post-processing. In this paper, we propose an entity resolution method for apartment properties using neural networks with inputs containing traditional property attributes as well as new attributes obtained from phonetic and semantic pre-processing of building names. The experimental results show that our proposed method improves entity resolution accuracy.