Examining causes of disputes in subcontracting litigation cases using text mining and natural language processing techniques

Disputes are common and widespread in subcontracting practices and would ultimately lead to litigation resulting in significant negative consequences including delay, increased cost, and tarnished reputation of contracting parties. Although considerable studies have analyzed disputes in construction projects, research efforts still fall short in investigating causes of disputes in subcontracting practices. To bridge the knowledge gap, this paper investigated the causes of disputes in subcontracting practices by automatically examining 3150 litigation cases that are publicly available in China using text mining and NLP (natural language processing) techniques. Documents of litigation cases were presented into vectors by data preprocessing and vector presentation. Clustering analysis of the data was conducted by using an improved K-Means model, K-Means++. Finally, 37 causes of disputes were discovered in the subcontracting litigation cases. Results showed that variations due to unforeseen circumstances, labor service issues from subcontractors, disagreement on engineering quantities between subcontractors and other contracting parties, invalid signature of subcontractors for obtaining more payment, payment issues between subcontractors and clients or general contractors, contradictory on the return of performance bond, disagreement on the interest of late payment, and arrears issues suffered by subcontractors were the top eight significant causes of disputes in subcontracting practices in China. This study contributes by systematically exploring the causes of disputes in subcontracting practices using automated text analysis approaches instead of manual content analysis. The findings of this study can provide construction practitioners a deeper understanding of disputes in subcontracting and thereby helping them increase their capacity to handle disputes.