Topic Modelling and Sentiment Analysis of AI-driven Circular Procurement practices

Abstract

The construction industry is the biggest worldwide consumer of raw materials, producing the most waste among all industries, with the UK construction sector generating 62% of national waste. Organisations can address these challenges through Circular Procurement (CP), which implements Circular Economy (CE) principles by incorporating reuse and recycling and material efficiency into procurement processes. However, CP remains underexplored in construction scholarship and practice. The research uses sentiment analysis and topic modelling to explore the integration of Artificial Intelligence (AI) in CP discourse. The research included 23 peer-reviewed articles from Scopus and Google Scholar databases. The sentiment analysis showed a positive attitude toward AI procurement, but most of the content was neutral because it focused on description rather than emotional expression. The LDA model produced five thematic areas, including machine learning and circularity. The corpus shows that circular procurement appears in only 3% of the text, which indicates its early stage of development in this field. The results show that AI methods already simplify procurement and could be extended to embed CE principles more effectively in construction. The originality of this research lies in combining sentiment analysis with topic modelling to provide a dual perspective on both tone and latent themes of the academic discourse. The research value comes from its demonstration of limited academic studies on AI-driven CP and its presentation of how existing AI procurement tools can be modified to support circularity. The research addresses current knowledge deficiencies while creating fundamental elements for sustainable procurement methods in construction projects.