Kadek Suarjuna Batubulan, Nobuo Funabiki, Komang Candra Brata, I Nyoman Darma Kotama, Yan Watequlis Syaifudin, Wen-Chung Kao, Yi-Fang Lee
Currently, we are developing a map information collection tool to assist a pedestrian navigation system. It collects building names and room numbers using Optical Character Recognition (OCR), geolocation techniques, and web scraping to extract necessary data from websites. Unfortunately, collected data is often incomplete, unstructured, or outdated, which can reduce the overall accuracy and reliability. In this paper, we investigate the accuracy of collected data by this tool through comparisons with reference data obtained from Google Maps API, and the applicability of generative AI models in detecting and correcting errors. For evaluations, we examine classification-based metrics such as precision, recall, and F1-score to measure the consistency and correctness of the collected data. © 2025 Copyright held by the owner/author(s).
Department of Information and Communication Systems, Okayama University, Okayama University, Okayama, Okayama, Japan; Department of Information Technology, Politeknik Negeri Malang, East Java, Malang, Indonesia; Department of Electrical Engineering, National Taiwan Normal University, Taipei, Taiwan; Department of Industrial Education, National Taiwan Normal University, Taipei, Taiwan
Research at a Glance
Register to unlock
Topics & SDG Alignment
Register to unlock
Collaboration
Register to unlock
Author Profile (Selected)
Register to unlock
References Overview
Register to unlock
Journal & Source
Register to unlock
Metadata & Integrity
Register to unlock