Skip to content 
GrandhooFishfort
A Data Analysis System Based on Logical Rules + Machine Learning
Vision
Simplify Data Analysis to Maximize Enterprise Data Value.
Theory
Graph Association Rules (GAR) Technology, Explainable AI Prediction, Automatic Discovery of Association in Data.
Technology
Provide automated intelligent data analysis methods, achieving high-precision, interpretable data analysis, data prediction, and data-driven decision-making.
Digital transformation requires better AI technology and associated data
Challenge 1：
Traditional Expert Rules: High Cost, Low Accuracy and Consistency.
Challenge 2：
AI Black Box Models: Results are Uninterpretable and Unreliable.
Challenge 3：
Balancing the contradiction between computational resources and efficiency in big data distributed computing. Is it possible to find new ways to reduce the total overhead of resources (communication, computation)?
Logic Rules + Machine Learning
Distributed graph association rules discovering and prediction
Parallel scalability based on different application
Incremental Graph Data Computation, Timeliness deduction
Explainable
Based on graph theory, it deduces logical rules and can explain the prediction results.
Rule Discovery
Continuously discovering and accumulating 'surprise' rules, discovering clues that ordinary people cannot see
Real-time response
It processes billions of data in seconds and supports more dimensions and large data ranges.
Security
Based on data features, it performs logical calculations, and user privacy data can still participate in calculations after being encrypted.
Key Technologies
Graph computing
It takes graph as data model and utilizes distributed parallel computing to explore the association rules existing in graph data.
Distributed computing
It supports real-time monitoring of node status information to ensure system reliability.
Logical rules and machine learning
It solves problems such as the accuracy of manually developed rules and improves the interpretability of machine learning results.
Automatic Feature Engineering
Perform data validation, cleaning, and feature extraction on customer data to obtain better data features.
Data Source Management
Provides the feature of converting relational data into graph data and centralized management of different data sources. The extracted graph data establishes a basis for rule discovery.
Data Access
Supports integration with multiple mainstream RDB system and allows importing CSV files.
Graph Configuration
It supports both intelligent graph construction and manual graph construction. Intelligent graph construction can automatically identify the relationships between various relational tables, reducing the cost of graph construction. Manual graph construction is simple, user-friendly, and provides an interactive experience.
Data Roles
Different data types play different roles in the data, and targeted processing is performed based on the roles to improve the efficiency and accuracy of the algorithm.
Model Configuration
Achieve the automatic extraction of business data attributes through feature engineering, enriching the attributes of data source, and enhance the accuracy of data analysis.
Rule Discovery
Provides graph data analysis and rule discovery services. Controls the granularity of mining by specifying data sources, support percentages, discovery depth, and configuration pattern structures. The system automatically generates rules, calculates rule evaluation metrics, retrieves instances, and offers a visual interface .
Lightweight Configuration
No coding required, achieve automatic discovery and analysis of hidden rules in data.
Visual Rules
View and understand data rules through visualization techniques, breaking the limitations of uninterpretable model analysis results.
Data Access
Supports integration with multiple mainstream RDB system and allows importing CSV files.
Graph Configuration
It supports both intelligent graph construction and manual graph construction. Intelligent graph construction can automatically identify the relationships between various relational tables, reducing the cost of graph construction. Manual graph construction is simple, user-friendly, and provides an interactive experience.
Data Roles
Different data types play different roles in the data, and targeted processing is performed based on the roles to improve the efficiency and accuracy of the algorithm.
Model Configuration
Achieve the automatic extraction of business data attributes through feature engineering, enriching the attributes of data source, and enhance the accuracy of data analysis.
Lightweight Configuration
No coding required, achieve automatic discovery and analysis of hidden rules in data.
Visual Rules
View and understand data rules through visualization techniques, breaking the limitations of uninterpretable model analysis results.
Item 1 of 2