Conversational Interface with Introspection for Expert Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing expert systems lack the ability to engage in real dialogue with users, provide explanations, enrich rule bases, or allow user interaction choices, limiting their effectiveness in controlling equipment and decision-making processes.
Innovation Solution
A computer-implemented device comprising a conversational interface and a knowledge-based system with fuzzy logic rule bases, which includes an extraction module to extract decision inputs, an introspection module to determine missing inputs, and a reconstruction module to provide natural language explanations and updates to the rule bases.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If traditional query interfaces with sentence models and blanks are used, then the system can generate responses based on calculated elements, but the system cannot engage in real dialogue or provide explanations to users
Solution Approach 1:
The patent introduces a natural language processing intermediary layer between the user and the knowledge-based system. This intermediary translates user questions into structured queries, processes responses through the expert system, and generates natural language explanations back to users, enabling both dialogue engagement and explanation provision simultaneously
Solution Approach 2:
The system implements feedback mechanisms where the expert system not only provides decisions but also generates explanations for those decisions. The system feeds back reasoning traces, rule activations, and knowledge base queries to users in natural language, allowing users to understand the basis of system recommendations and engage in follow-up dialogue
2Productivity
If fixed rule bases are used in expert systems, then the system can make automated decisions, but the system cannot enrich its rule bases or adapt to new interactions
Solution Approach 1:
The patent implements self-service mechanisms where the expert system automatically learns from user interactions, validates new rules against existing knowledge bases, and enriches its rule base without requiring manual programming. The system autonomously identifies patterns in user queries, generates candidate rules, and integrates them into the knowledge base, maintaining automated decision capability while enabling continuous adaptation
Solution Approach 2:
The rule base is transformed from a static structure to a dynamic one that evolves through user interactions. The system dynamically adds, modifies, and validates rules based on encountered scenarios, allowing the expert system to adapt its decision-making capabilities while maintaining automated operation
3Ease of operation
If natural language processing is implemented for conversational interface, then user interaction becomes more intuitive, but the system complexity increases
Solution Approach 1:
The natural language processing system is segmented into distinct functional modules: question parsing module, intent recognition module, query generation module, response interpretation module, and explanation generation module. Each module handles a specific aspect of language processing, reducing overall system complexity while enabling comprehensive natural language interaction capabilities
Data Source
Figure 1
Figure 2
Figure 3~4A
AI summary
The invention relates to a computer-implemented device comprising a conversational interface and a knowledge-based system, the knowledge-based system comprising fuzzy logic rule bases (7) and the conversational interface comprising: - an extraction module (11) configured to extract from a question relating to a search for a decision, corresponding decision search inputs which are adapted to be used by said knowledge-based system (5) in order to generate a decision to said decision inputs, - an introspection module (13) configured in case said decision search inputs are insufficient, to determine additional decision search inputs which are missing for the knowledge-based system to generate the decision,the introspection module (13) being configured to determine said additional decision-making search inputs by searching among all the fuzzy logic rules those which could not be triggered and the reason for this non-triggering, and - a reconstitution module (15) configured to reconstitute said additional decision-making search inputs from the introspection module (13) or the decision from the knowledge-based system (5) into a natural language text.,