Deep Query Parsing Model for AI Dialog Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current human-machine dialog products face challenges in accurately parsing spoken queries due to the diversity and complexity of speech expressions, as they rely on limited, manually summarized templates that fail to correctly interpret more complex queries.
Innovation Solution
A method and apparatus using artificial intelligence to parse queries by obtaining a knowledge library specific to an application domain, training a deep query parsing model, and employing word segmentation and embedding representations to determine intent and parameter information, with a CRF model for global slot notation, allowing for improved parsing accuracy across various query representations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If manually summarized templates are used for query parsing, then the system can be simple to implement, but the parsing accuracy deteriorates due to limited template coverage
Solution Approach 1:
The patent replaces the manual template-matching mechanism with a deep learning-based automatic parsing system. The deep query parsing model learns parsing patterns automatically from training data rather than relying on manually crafted templates, thereby improving parsing accuracy while maintaining implementation feasibility through automated model training.
Solution Approach 2:
The patent changes the fundamental parameter of the parsing system from fixed manual templates to dynamic learned representations. By using word embedding and deep neural networks, the system adapts its parsing behavior based on learned parameters from training data, enabling it to handle diverse query expressions beyond predefined templates.
2Adaptability or versatility
If diverse speech expressions are supported, then the system becomes more versatile, but the complexity of understanding queries increases
Solution Approach 1:
The patent creates a universal deep query parsing model that handles multiple application domains and diverse speech expressions through a single unified system. The model learns domain-specific knowledge from training data and can generalize to handle various query types without requiring separate processing mechanisms for each domain or expression style.
Solution Approach 2:
The system employs self-service mechanisms where the deep learning model automatically learns parsing strategies from training data without manual intervention. The model performs self-adjustment during training to handle diverse expressions, reducing the need for complex manual configuration and rule-setting for each new expression type.
3Use of energy by moving object
If manual template matching is used, then the system requires less computational resources, but it cannot correctly parse complex queries
Solution Approach 1:
The patent performs preliminary action by pre-training the deep query parsing model on extensive training data before deployment. This pre-learning phase enables the model to internalize parsing patterns and relationships, allowing it to correctly parse complex queries during actual use without requiring excessive computational resources at runtime compared to manual template matching.
Data Source
AI summary
The present disclosure provides a method and apparatus for parsing a query based on artificial intelligence, and a storage medium, wherein the method comprises: regarding any application domain, obtaining a knowledge library corresponding to the application domain; determining a training query serving as a training language material according to the knowledge library; obtaining a deep query parsing model by training according to the training language material; using the deep query parsing model to parse the user's query to obtain a parsing result. The solution of the present disclosure may be applied to improve the accuracy of the parsing result.