An AI-based Approach for Mystery Shopping Audit in Customer Service

機構典藏 > College of Engineering > Graduate Institute & Department of Computer Science and Information Engineering > Proceeding > Item 987654321/126361

Please use this identifier to cite or link to this item: https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/126361

Title:	An AI-based Approach for Mystery Shopping Audit in Customer Service
Authors:	Chuang, Christopher;Zhang, Qiaoyun;Lin, Yi-Ti;Ho, Chia-Ling;Chang, Chih-Yung
Date:	2024-05-22
Issue Date:	2024-10-07 12:06:07 (UTC+8)
Publisher:	Association for Computing Machinery,New York,NY,United States
Abstract:	In the era of intense business competition, the emergence of mystery shoppers provides companies with objective insights, enabling them to innovate and enhance their offerings to meet evolving customer needs and maintain a competitive edge. This paper introduces AMSA (AI-based approach for Mystery Shopping Audit), a service behavior identification mechanism for analyzing mystery shopping audit videos. AMSA identifies behaviors like five-finger guidance, hand offering, and maintaining good posture through two phases: coarse-grain and fine-grain identification. In the coarse-grain phase, an automated filtering and classification algorithm is proposed, utilizing YOLO for target detection. Subsequently, fine-grain identification employs 3DCNN for action classification, trained on enhanced videos of target actions. The Simulation results show that the proposed AMSA significantly improves accuracy of identification.
DOI:	10.1145/3658549.3658555
Appears in Collections:	[Graduate Institute & Department of Computer Science and Information Engineering] Proceeding

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	74	View/Open