A Novel Approach for Company Real Workplace Identification via E-commercial Data

Urban growth benefits significantly from local business development. However, factors like traffic and labor shortages sometimes cause companies to operate away from their registered addresses, resulting in governance challenges. This paper introduces "LocRecognizer,"a data mining method that leverages e-commerce data to pinpoint companies' real-world operational locations. Based on the principle that areas with a high concentration of company-related users likely indicate actual workplaces, LocRecognizer combines hierarchical clustering with a deep learning model for accurate detection. When tested on datasets from Beijing and Nantong, it outperformed six baselines. A practical implementation of this system has been operational in Nantong since September 2021, attesting to its effectiveness.