Intent Determination via Coherent Subgroup Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current conversation processing devices face challenges in accurately determining user intent from text due to ambiguity and context dependence, relying on human-created ontologies and semantic grammars, which can be limited and prone to misinterpretation.
Innovation Solution
A system that generates distinct intent labels, divides them into interpretation partitions with overlapping semantic content, defines maximal coherent subgroups, computes scores, and selects the best subgroup to aggregate intent labels, providing a multiple intent interpretation of the conversation element.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If human-created ontologies and semantic grammars are used for intent determination, then the system can operate with existing knowledge structures, but the accuracy and robustness of intent determination deteriorates due to ambiguity and context dependence
Solution Approach 1:
The patent segments the intent determination process into multiple independent modules: a natural language understanding module that generates intent labels, a semantic graph module that builds relationships between concepts, and a classification module that determines user intent. This segmentation allows each module to specialize in specific tasks, improving overall accuracy while maintaining implementation feasibility through modular architecture.
Solution Approach 2:
The patent introduces a semantic graph as an intermediary structure between the input text and the final intent classification. The semantic graph acts as a mediator that captures semantic relationships, ambiguity, and context information, transforming raw text into structured semantic representations that improve intent determination accuracy without requiring complete redesign of the entire system.
2Device complexity
If traditional intent determination methods are used, then the system structure remains simple, but the ability to handle ambiguity and context dependence deteriorates
Solution Approach 1:
The patent implements a nested architecture where the semantic graph module is embedded within the natural language understanding module, which in turn is integrated with the classification module. The semantic graph contains nested semantic graphs for different concepts, and intent labels are nested within semantic relationships. This nesting allows complex ambiguity and context handling to be contained within modular structures, maintaining overall system simplicity while improving reliability.
3Loss of time
If human-created ontologies are used, then development time is reduced, but the adaptability to new contexts and ambiguities deteriorates
Solution Approach 1:
The patent implements a dynamic semantic graph that can automatically adapt to new contexts and ambiguities through machine learning and user feedback mechanisms. The system dynamically updates semantic relationships and intent classifications based on real-world usage patterns, allowing it to adapt to emerging contexts without requiring manual ontology updates. This dynamic approach maintains development time efficiency while significantly improving adaptability.
Data Source
AI summary
A system determines user intent from a received conversation element. A plurality of distinct intent labels are generated for the received conversation element. The generated plurality of distinct intent labels are divided into a plurality of interpretation partitions with overlapping semantic content. for each interpretation partition of the plurality of interpretation partitions, a set of maximal coherent subgroups are defined that do not disagree on labels for terms in each subgroup, a score is computed for each maximal coherent subgroup of the defined set of maximal coherent subgroups, and a maximal coherent subgroup is selected from the set of maximal coherent subgroups based on the computed score. Intent labels are aggregated from the selected maximal coherent subgroup of each interpretation partition of the plurality of interpretation partitions to define a multiple intent interpretation of the received conversation element. The defined multiple intent interpretation is output for the received conversation element.