Procurement and cost-control processes in the facility management sector rely on the manual evaluation of supplier quotations, a practice that does not scale with volume and is inherently subjective: the quality of each assessment depends on the individual reviewer’s experience, and no systematic benchmarking methodology is in place. The problem is compounded by the semantic heterogeneity of quotation documents, in which suppliers describe the same work item using different terminology, making direct com- parison against standard reference price lists unreliable. This thesis presents the design, implementation and validation of an automated quo- tation analysis system developed in collaboration with Sprint Reply within an industrial procurement context.