Construction Knowledge Graph Risk Scoring for Project Delays
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Construction projects face inefficiencies and risks due to the siloed nature of project information, leading to difficulties in visualizing interrelationships, data loss, and unpredictable delays, which existing solutions like common data environments fail to address.
Innovation Solution
A construction knowledge graph utilizing machine-learning techniques to connect data objects based on unifying attributes, such as location, and apply risk quantification to identify and mitigate risks through machine-learning models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional construction project management methods are used to handle vast amounts of information, then project coordination and stakeholder communication are maintained through conventional means, but data connectivity between different data types and stakeholders is insufficient, leading to increased risks and delays
Solution Approach 1:
The patent merges multiple data types (drawings, specifications, schedules, contracts) and stakeholder information into a unified knowledge graph structure. This integration enables comprehensive risk assessment by connecting previously siloed information sources, allowing the system to evaluate risks holistically across the entire construction project rather than in isolated segments.
Solution Approach 2:
The knowledge graph acts as an intermediary layer between raw construction data and risk assessment outcomes. It transforms unconnected data objects into a network of relationships, enabling the machine learning models to infer risk scores by analyzing patterns across the interconnected data structure, thereby bridging the gap between data collection and risk management.
2Measurement precision
If machine-learning models are trained using only direct data object attributes, then training complexity is reduced, but the models fail to capture indirect risk relationships and cascading effects between connected data objects
Solution Approach 1:
The patent transitions from analyzing data objects in isolation (one-dimensional) to analyzing them within a multi-dimensional knowledge graph structure. By incorporating the degree of separation and connectivity patterns as additional dimensions, the model captures indirect relationships and cascading effects while maintaining a structured approach that manages training complexity through graph-based feature extraction.
Solution Approach 2:
The training process is segmented into distinct stages: first, the knowledge graph is constructed and pre-processed; second, features are extracted based on connectivity patterns and degree of separation; third, the machine learning model is trained on these structured features. This segmentation allows complex risk relationship modeling to be broken down into manageable computational steps.
3Reliability
If comprehensive risk assessment considering all connected data objects is performed, then risk identification accuracy is improved, but computational time and processing resources increase significantly
Solution Approach 1:
The patent applies local quality by focusing computational resources on evaluating data objects and relationships based on their specific connectivity patterns and risk relevance. Rather than uniformly processing all data objects, the system prioritizes analysis of highly connected nodes and critical path relationships, allocating more computational effort to areas with higher risk impact while reducing effort for less critical connections.
Solution Approach 2:
The system performs partial action by limiting the depth of traversal in the knowledge graph to a maximum degree of separation. Instead of exhaustively analyzing all possible indirect relationships across the entire graph, the model evaluates risks up to a predetermined connectivity threshold, capturing the most significant indirect effects while avoiding the computational burden of complete graph analysis.
4Productivity
If real-time risk scores and actionable recommendations are provided to stakeholders, then project efficiency and risk mitigation are improved, but the complexity of the system architecture and implementation increases
Solution Approach 1:
The knowledge graph and machine learning models operate autonomously to continuously assess risks and generate recommendations without requiring manual intervention. The system self-updates as new data objects are added or modified in the construction project, automatically recalculating risk scores and providing real-time guidance to stakeholders, thereby reducing operational complexity despite the sophisticated underlying architecture.
Data Source
AI summary
A computing platform is configured to (i) receive data objects related to a construction project, (ii) add the data objects to a construction knowledge graph as nodes that are connected to other nodes representing other data objects, (iii) determine, via a machine-learning model trained using historic construction project data, a first risk score for a first data object, (iv) determine, via the machine-learning model, a second risk score for a second data object, where the second risk score is based on (a) the first risk score and (b) a degree of separation between the first data object and the second data object in the construction knowledge graph, (v) based on the second risk score, automatically generate a suggested action to be taken with respect to the first data object, and (vi) cause an indication of the suggested action to be displayed at a client station of a user associated with the construction project.