Extracting Construction Knowledge from Project Schedules Using Natural Language Processing

A sound and good quality schedule is critical to the success of a construction project. However, the little time available for proper project scheduling in the planning and design stage often impairs the quality of a schedule. Few efforts have been made to evaluate and maintain the schedule quality in the construction stage. Usually project teams need to put intensive manual efforts to conduct schedule quality diagnosis which is time-consuming and subjective to a large extent. One major challenge of diagnosing schedule quality is understanding the activity characteristics and construction logic. The multi-partite nature of construction projects (i.e. schedulers and project teams) further exacerbates the difficulty of diagnosis. This paper thus proposes a novel semantic-based logic reasoning and representation methodology to extract construction methods from the schedule to ensure a consistent project schedule. The intellectual contributions of this paper are twofold. First, this paper develops an ontology of tasks with hierarchies from the schedule to automatically extract the construction methods and activities. Second, this paper presents a novel dependency-based information representation schema for representing the logics between tasks and key constraints to facilitate the complete automation in construction logic reasoning from the schedule. To test the proposed system, this paper evaluates the average rate of recall and precision achieved by the system for extracting construction activities and logics in the schedule within one month and compared the results with the rate achieved by manual check. The developed system provides both academics and practitioners a method to detect the deficiencies of project schedules and assists project planners to produce and maintain good quality schedules starting from project initiation until its completion.