Hybrid AI Action Recommendation System
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computer systems face challenges in identifying an optimal set of actions to take in response to events, as certain actions may not effectively resolve the issue and may be disfavored by users.
Innovation Solution
A hybrid artificial intelligence system processes events to generate actionable recommendations by filtering action sets using objective values and probabilities, optimizing them based on user preferences and rule data to select the most effective actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a computer system identifies events and performs actions in response, then the system can respond to changes in data, but it faces challenges in identifying an optimal set of actions to take
Solution Approach 1:
The patent introduces an intermediary optimization service that acts as a mediator between the event detection system and the action execution system. This service receives events, generates candidate actions using multiple AI models (LLM, traditional ML models), filters them through rule-based systems, and optimizes the final action set. This intermediary layer resolves the contradiction by handling the complex action selection process centrally, improving response efficiency while managing complexity through a dedicated optimization component rather than distributed decision-making.
Solution Approach 2:
The action selection process is segmented into distinct stages: event processing, candidate action generation by multiple specialized models, filtering using rule data and user preferences, and final optimization. Each segment handles a specific aspect of the problem, allowing the system to improve overall productivity by dividing the complex task into manageable parts that can be processed independently and in parallel where possible.
2Adaptability or versatility
If the system generates multiple candidate actions for an event, then it has more options to choose from, but it increases the difficulty of selecting the most effective actions
Solution Approach 1:
The system implements feedback mechanisms where the optimization service evaluates candidate actions based on multiple criteria including user preferences, rule data, and predicted effectiveness. The action models provide feedback scores for each candidate action, and the optimization process uses this feedback to rank and select the most effective actions. This feedback-driven approach allows the system to maintain high adaptability with multiple action options while systematically evaluating and selecting the best actions based on measurable effectiveness criteria.
Solution Approach 2:
The patent changes the parameters used to evaluate actions by introducing multiple dimensions: objective values from traditional ML models, probability scores, user preference weights, and rule-based constraints. By transforming the action evaluation from a single metric to a multi-parameter optimization problem, the system can handle diverse action options adaptably while using structured parameter changes to make the evaluation process systematic and measurable rather than subjective and difficult.
3Reliability
If the system uses traditional machine learning models to identify actions, then it can process events systematically, but it may not effectively capture complex user preferences and contextual nuances
Solution Approach 1:
The patent merges traditional machine learning models with large language model (LLM) based action models. The traditional ML models provide reliable, systematic event processing with proven performance on structured data, while the LLM-based models capture complex user preferences and contextual nuances through natural language understanding. By combining these different model types in a multi-model action generation system, the patent achieves both reliability from traditional methods and adaptability from AI-based methods, allowing the system to leverage the strengths of each approach.
Data Source
AI summary
A method implements hybrid artificial intelligence generated actionable recommendations. The method includes processing an event to identify an action of an event action set. The event includes an event value. The method further includes processing the event action set to generate an objective value, corresponding to the action, and a probability, corresponding to the action, and to form a model action set from the event action set. The method further includes filtering the model action set using action rule data and rule user data to generate a filtered action set. The method further includes processing, using the objective value and the probability, the filtered action set with an optimization controller to generate suggested action sets from which a selected action set is selected. The selected action set corresponds to a combined action value that satisfies the event value.