Context Carryover in Language Understanding Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current language understanding systems fail to accurately track user intent in multi-turn interactions by ignoring system responses and knowledge updates, leading to inaccurate context carryover and goal tracking.
Innovation Solution
A system that utilizes a common schema to identify entities in both natural language inputs and responses, tracks carryover entities across turns, and determines user intent by combining current and carryover entities, thereby enhancing contextual language understanding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If system responses and knowledge updates are ignored in multi-turn interactions, then the system complexity is reduced, but the accuracy of user intent tracking deteriorates
Solution Approach 1:
The system segments the conversation into discrete turns, each containing user inputs and system responses. By processing each turn as a segmented unit with structured entities, the system maintains manageable complexity while capturing complete interaction context for accurate intent tracking.
Solution Approach 2:
The system performs preliminary entity identification and schema mapping on both user inputs and system responses before intent determination. This preliminary processing of contextual information from previous turns enables accurate intent tracking without adding complexity during the main intent determination process.
2Loss of time
If only current natural language inputs are processed, then the processing time is reduced, but the reliability of context carryover deteriorates
Solution Approach 1:
The system performs preliminary entity extraction and schema validation on previous turns during idle time or asynchronously, so that when current intent determination is needed, the contextual data is already prepared and available for immediate use, minimizing processing time impact.
Solution Approach 2:
The system incorporates feedback from system responses and knowledge updates into the context carryover mechanism. By continuously updating and validating contextual entities based on actual system behavior and user interactions, the system maintains high reliability without requiring extensive real-time processing.
3Stability of the object's composition
If a common schema is applied to both natural language inputs and responses, then the consistency of entity identification is improved, but the device complexity increases
Solution Approach 1:
The system creates a universal schema that serves multiple functions: it structures user inputs, processes system responses, identifies entities, and maintains contextual carryover. This single multi-functional schema eliminates the need for separate processing mechanisms, improving consistency while managing complexity through consolidation.
Solution Approach 2:
The system uses schema templates that can be copied and reused across different conversation turns and contexts. By storing validated schema instances and reusing them, the system maintains consistent entity identification without requiring complex real-time schema generation or interpretation.
Data Source
AI summary
Systems and methods for determining a user intent or goal for contextual language understanding by utilizing information from one or more previous user natural language inputs and one or more previous system generated responses to the user natural language inputs are provided. More specifically, the systems and methods utilize a common schema for determining features from the responses and natural language inputs and provide carryover tracking between responses and the natural language inputs. Accordingly, the systems and methods for contextual language understanding provide for a more accurate, a more reliable, and a more efficient context carryover and goal tracking system when compared to systems and methods that do not utilized the responses in determining the user goal/intent.