Despite the ubiquitous presence of textual data in daily life and their significance for businesses, textual data have not been investigated proactively in the housing industry. The unstructured nature of textual data is a key obstacle. This study aims to address this gap by fully using text documents related to housing management and providing both residents and property managers with insights.
Using text vectorization methods, such as term frequency-inverse document frequency and word embeddings, 9,023 consultation records from the Seoul Support Center for Apartment Management were converted into numeric data. Subsequently, the numeric data were fed into a k-means clustering algorithm for document classification.
Eight distinct clusters were identified and analyzed. Each cluster represents a unique category: general inquiries, management regulations, vendor company selection, residents’ representative council, budgeting, interpretation of laws, long-term repair plans and the ministry responsible for apartment management.
The approach adopted in this study is expected to enhance housing management practices by facilitating the prompt classification of resident inquiries, thereby optimizing housing policies and practice.
