Learn how OpenRules Machine Learning and Decisioning tools can help you build constantly learning decision-making applications
The modern field of Machine Learning offers powerful algorithms for the extraction of patterns from large collections of historical data to present them in the form of readable classification rules.
Rule Learner is an open source tool that naturally integrates Machine Learning (ML) and Business Rules (BR) techniques by incorporating ML algorithms into rules-based Decision Models.
Videos:
ML+BR Integration
Analytical World. In the analytical world, business analysts take Historical Data from the Enterprise Data Repository and apply their expert knowledge to extract records (“Training Instances”) which can be used to learn classification rules. It can be done manually or using a rule engine (called “Rule Trainer“) that applies Training Rules created by business analysts to the historical records. Business analysts then run Rule Learner against the resulting training instances by experimenting with different ML algorithms until Rule Learner produces adequate classification rules. The generated rules will be placed into the Enterprise Rules Repository.
Operational World. Being incorporated into business decision models, these rules will be used in the operational world to help to make the correct decisions using real-time data streams. The generated rules will be executed by a rule/decision engine that will produce new decisions. New data records, saved back into the Enterprise Data Repository, will serve as a source for further adjustments in the automatically generated business rules.
Ever-Learning Loop. The described ML and BR integration puts a decision-making system into an ever-learning loop that supports self-learning business processes. This way Rule Learner becomes a good “employee” of the enterprise that can learn from the previous experiences and produce new knowledge. Rule Learner and Rule Engines working together enrich the decisioning system with self-improving learning capabilities!
Implementation Notes. Rule Learner has been designed as an ML component of the Digital Decisioning System “OpenRules Decision Manager“. However, being an open source product, it also can be incorporated into other digital decisioning systems. Rule Learner is written in Java and comes with the source code. It provides a simple API for incorporation of Rule Learner in any on-premise or on-cloud software infrastructure.
Rule Learner is oriented toward business analysts by allowing them to present their historical data in simple Excel tables and then use that data to automatically generate business rules and incorporate them into executable business decision models. It can also work with data presented in JSON or XML. Rule Learner is easy to install and to use without becoming an expert in data science or programming.
