The increasing complexity of design coordination in Building Information Modeling (BIM) highlights the need for intelligent tools that support issue resolution. While the BIM Collaboration Format (BCF) facilitates issue reporting, it lacks mechanisms for learning from past experiences to inform the resolution of new issues. This thesis presents a Case-Based Reasoning (CBR) framework that integrates BCF issue topics with Industry Foundation Classes (IFC) model data to enable the retrieval and reuse of similar past coordination issues. This information from each BCF topic and its associated IFC elements is transformed into structured JSON cases. The system includes a multi-dimensional similarity method, based on textual embeddings, geometry, and connectivity. Resolution suggestions from retrieved cases are enhanced using a language model, which rewrites vague descriptions into clearer, actionable recommendations. The system was evaluated in two phases: one using a homogeneous dataset and another with more diverse inputs. Results show over 60% top-1 retrieval accuracy and a 70% acceptance rate for proposed resolutions. These findings demonstrate the value of combining CBR and language models to support BIM coordination by reducing cognitive load and improving resolution consistency. The work lays a foundation for integrating knowledge-driven support into BIM workflows and suggests paths for future enhancement.     «

The increasing complexity of design coordination in Building Information Modeling (BIM) highlights the need for intelligent tools that support issue resolution. While the BIM Collaboration Format (BCF) facilitates issue reporting, it lacks mechanisms for learning from past experiences to inform the resolution of new issues. This thesis presents a Case-Based Reasoning (CBR) framework that integrates BCF issue topics with Industry Foundation Classes (IFC) model data to enable the retrieval and reu...     »