NLU Query Processing with Obsequious Expression Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current voice recognition systems face inaccuracies when dealing with soft-spoken or unclear inputs, particularly in children, and require additional training and complex architecture to handle polite expressions, leading to reduced accuracy and user frustration.
Innovation Solution
A natural language model-based voice recognition system that minimally trains and adapts for legacy systems, using a binary classifier to distinguish between queries with and without obsequious expressions, allowing for plug-and-play modularity and improved recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If voice recognition systems require polite expressions (please, thank you) to be included in commands, then user politeness is encouraged, but recognition accuracy decreases and match scores are reduced
Solution Approach 1:
The system segments the command processing into two independent parts: (1) a politeness detection module that identifies polite expressions in the input, and (2) a command recognition module that processes the core intent. By separating these functions, the system can acknowledge politeness without letting it interfere with accurate command recognition and matching.
Solution Approach 2:
The patent introduces an intermediary politeness detection layer that acts as a mediator between the user's polite expression and the command recognition system. This intermediary identifies and isolates polite expressions (like 'please' and 'thank you') before the core command is processed, preventing these expressions from degrading the match scores while still encouraging user politeness.
2Adaptability or versatility
If additional training data and model complexity are added to handle polite expressions, then the system can recognize polite commands, but training time and computational resources increase significantly
Solution Approach 1:
The system divides the training process into separate modules: one for training the core command recognition model and another for training the politeness detection component. This segmentation allows each module to be trained independently with targeted data, reducing the overall training time and computational requirements compared to training a single monolithic model to handle both functions.
Solution Approach 2:
Instead of training the entire system to recognize and process all variations of polite expressions, the patent implements a focused approach that detects common polite expressions (please, thank you) using a lightweight separate module. This partial action approach provides sufficient adaptability for polite expressions without the excessive training overhead of a comprehensive solution.
3Reliability
If obsequious expressions are removed from search queries to improve matching, then legacy system compatibility is maintained, but queries with entity titles containing such expressions yield incorrect results
Solution Approach 1:
The patent introduces an intermediary detection layer that identifies whether an expression like 'please' or 'thank you' is a genuine polite expression or part of the actual search query (entity title). This intermediary analysis allows the system to maintain legacy compatibility by removing polite expressions when appropriate while preserving search accuracy when the expression is part of the query intent.
Solution Approach 2:
The system applies different processing rules to different parts of the query based on local context. The politeness detection module analyzes the specific position and context of expressions within the query, removing polite expressions only when they are standalone modifiers rather than integral parts of entity titles or search terms, thereby maintaining both legacy compatibility and search accuracy.
Data Source
AI summary
Systems and methods for generating a query using a trained natural language model of a natural language understanding (NLU) system are disclosed herein. A text string including at least a content entity is received. A determination is made as to whether the text string includes an obsequious expression. In response to determining the text string includes an obsequious expression, a determination is made as to whether the obsequious expression describes the content entity. In response to determining whether the obsequious expression describes the content entity, the query is generated. In response to determining the obsequious expression describes the content entity, the content entity and the obsequious expression are included in the query and in response to determining the obsequious expression does not describe the content entity, the content entity is included in the query and the obsequious expression is excluded from the query.