Contextual Search Query Revision via Candidate N-gram Probability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current natural language processing systems for e-commerce transactions lack contextual recall, providing unnatural responses and failing to reference prior utterances for identifying known information, leading to robotic conversations and limited capabilities within 'walled gardens' of specific software and hardware platforms.
Innovation Solution
A system and method that generate candidate n-grams from user utterances, including prior semantic components, to calculate probabilities and output intended responses, enabling contextual understanding and multi-turn conversations across various platforms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If current natural language systems are used to provide responses to general queries, then the system operation is simple, but the contextual recall capability is limited and conversations sound robotic
Solution Approach 1:
The system performs preliminary actions by generating candidate n-grams from prior semantic components before the user provides their utterance. This allows the system to proactively prepare contextual information and probability calculations, improving contextual recall without requiring complex real-time processing during conversation
Solution Approach 2:
The system segments the contextual understanding task into distinct components: extracting semantic components from prior utterances, generating candidate n-grams, calculating probabilities for each candidate, and selecting the intended n-gram. This segmentation allows each component to be processed independently, managing system complexity while improving contextual recall
2Measurement precision
If the system generates multiple candidate n-grams with probability calculations, then the contextual understanding improves, but the processing time increases
Solution Approach 1:
The system applies partial action by generating a limited set of candidate n-grams rather than exhaustively processing all possible combinations. By focusing on the most probable candidates derived from prior semantic components, the system achieves sufficient contextual understanding accuracy without excessive processing time
Solution Approach 2:
The system replaces traditional mechanical sequence matching with a probability-based selection mechanism. Instead of rigidly comparing utterances against predefined sequences, the system calculates probabilities for candidate n-grams and selects the most likely intended meaning, improving accuracy while reducing processing overhead through statistical methods
3Adaptability or versatility
If digital assistants operate in walled gardens with identical software and hardware, then the system stability is maintained, but the adaptability across different platforms is limited
Solution Approach 1:
The system achieves universality by designing a platform-agnostic natural language processing architecture that can extract semantic components and generate candidate n-grams regardless of the underlying software or hardware platform. This allows the same core processing logic to adapt to different platforms while maintaining consistent contextual understanding capabilities
Solution Approach 2:
The system introduces an intermediary layer of semantic component extraction and n-gram generation that sits between the user input and the platform-specific response generation. This intermediary layer translates diverse platform inputs into a unified semantic representation, enabling cross-platform adaptability while preserving system stability through consistent processing rules
Data Source
AI summary
Systems and methods for contextual search query revision are disclosed. A user utterance including at least one semantic component is received and a plurality of candidate n-grams including the at least one semantic component and at least one additional semantic component selected from a set of prior semantic components is generated. A probability that each of the plurality of candidate n-grams is an intended n-gram is calculated and a selected one of the plurality of candidate n-grams is output based on the probability.