Fuzzy Cognitive Map Prediction Using Linguistic Terms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing fuzzy cognitive map (FCM) approaches lack the interrelated features necessary for accurate forecasting and prediction of future events, as they fail to effectively iterate and converge node states based on exogenous and non-exogenous inputs, leading to imprecise predictions.
Innovation Solution
A computer-implemented method and system that iteratively aggregates non-exogenous nodes in an FCM based on the states of exogenous and other non-exogenous nodes, using linguistic terms to represent causal relationships and iterate the map to a convergence point for generating predictions, allowing for real-time and dynamic adjustments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If FCM approaches use traditional numerical methods to compute node states, then calculation speed is improved, but interpretability and realism of predictions deteriorate
Solution Approach 1:
The patent replaces traditional numerical computation methods with a linguistic-based computation system. Instead of using mathematical equations and numerical values to represent node states and causal relationships, the system uses linguistic terms (words) from predefined vocabularies. This substitution allows the system to maintain computational efficiency while producing predictions that are directly interpretable in natural language, thus resolving the contradiction between calculation speed and interpretability.
Solution Approach 2:
The patent fundamentally changes the parameter representation from numerical values to linguistic terms. Node states are represented by words from vocabularies rather than numerical activations, and causal relationships are represented by linguistic strength values rather than numerical weights. This parameter transformation enables the system to compute predictions quickly while maintaining high interpretability, as the output predictions are expressed in the same linguistic framework used throughout the computation.
2Measurement precision
If FCM approaches use detailed numerical computations to achieve precise predictions, then measurement precision is improved, but ease of operation and interpretability worsen
Solution Approach 1:
The patent substitutes complex numerical computation mechanisms with a linguistic-based system. The aggregation functions that traditionally operate on numerical values are replaced with linguistic aggregation operations that work directly with words from predefined vocabularies. This substitution maintains prediction precision while dramatically improving ease of operation and interpretability, as users can understand and work with linguistic terms without requiring expertise in numerical methods or mathematical models.
3Reliability
If FCM approaches manually create cognitive maps with subject matter experts, then reliability of concept extraction is improved, but productivity and time consumption worsen
Solution Approach 1:
The patent enables the cognitive map creation process to be partially automated through self-service mechanisms. The system can automatically extract concepts and causal relationships from text corpora using natural language processing techniques, reducing the manual effort required while maintaining reliability. Subject matter experts can review and refine the automatically extracted maps, combining the speed of automated extraction with the reliability of expert validation, thus resolving the contradiction between reliability and productivity.
Solution Approach 2:
The patent introduces natural language processing algorithms as an intermediary between raw text data and the cognitive map structure. This intermediary automatically processes large volumes of text to extract concepts and relationships, providing a bridge that maintains the reliability of expert-created maps while dramatically increasing productivity. The NLP intermediary processes text corpora to generate initial cognitive maps that can then be refined by subject matter experts, combining automated efficiency with expert reliability.
Data Source
AI summary
A program product comprising; formulating a map of a plurality of nodes; establishing a set of causal relationships between the plurality of nodes, creating a connection between the nodes where a causal relationship is determined, and assigning a linguistic term describing the connection strength to each such connection; manipulating the plurality of nodes based on the set of exogenous nodes and a set of non-exogenous nodes; repositioning the plurality of nodes in the map; iterating the map until a convergence state is reached for each non-exogenous node; assigning a linguistic term to the relationships between the nodes based on a relationship between word index values and their corresponding linguistic terms; calculating the states of the at least one non-exogenous node based on the linguistic terms associated with all of the connected nodes; and generating a graphical representation in a user interface of the state of the selected non-exogenous node.