Resolving inconsistency in building information using uncertain knowledge graphs: a case of building space management

Abstract: The challenge of inconsistent and conflicting information in building space management is addressed using the uncertain knowledge graph. In this method, facts are represented in first-order logic formulas with weights attached to quantify credibility. Conflict-free information is acquired through the Maximum A-Posterior query over the Markov logic network. Semi-structured interviews with asset management professionals helped understand the general space management process in commercial buildings and its challenge of conflicting information. A case study showed that trusted space information can be drawn from dissimilar sources by eliminating redundant and inconsistent information, hence, reducing expensive manual verification processes.