Topic Determination via Template Similarity Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing dialogue systems rely heavily on supervised learning methods that require manual annotation of large training corpora, leading to high labor costs and low accuracy in topic recognition due to varying annotation standards.
Innovation Solution
A method and apparatus that determine a topic by calculating similarities between a to-be-recognized sentence sequence and pre-set topic templates, reducing the need for extensive manual annotation and improving recognition accuracy by using a server with processors and storage to implement the method.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If supervised learning method with manual annotation is used, then topic recognition model can be trained, but labor costs increase and annotation accuracy decreases due to varying standards
Solution Approach 1:
The patent creates topic templates that serve as standardized copies of topic structures. Instead of manually annotating each dialogue, the system uses pre-defined topic templates (containing topic names, keywords, and sentence patterns) to automatically match and recognize topics in dialogues, eliminating the need for repetitive manual annotation while maintaining consistent recognition standards
Solution Approach 2:
The patent transforms the topic recognition approach from manual classification to automated parameter-based matching. By defining topics through structured templates with specific parameters (topic name, keywords, sentence patterns, entity types), the system changes the recognition mechanism from human judgment to automated template matching, improving both efficiency and consistency
2Reliability
If manual annotation of large training corpus is performed, then supervised learning model can be built, but labor costs increase significantly
Solution Approach 1:
The patent performs preliminary action by pre-defining topic templates before actual topic recognition. The templates include pre-established topic names, keywords, sentence patterns, and entity type mappings. This preliminary structuring allows rapid automated matching during recognition, eliminating the time-consuming manual annotation process while maintaining high accuracy
Solution Approach 2:
Instead of creating new annotations for each training case, the system copies and reuses standardized topic templates across multiple dialogues. This template-based approach allows the same topic structure to be applied repeatedly without requiring new manual annotations, dramatically reducing annotation time and labor costs
Data Source
AI summary
Embodiments of the present disclosure disclose a method and apparatus for determining a topic. A specific embodiment of the method comprises: determining a to-be-recognized sentence sequence; calculating similarities between the to-be-recognized sentence sequence and each of topic templates in a topic template set in a target area, the each of the topic templates in the topic template set corresponding to a topic in at least one topic in the target area, the topic template including a topic section sequence, and a topic section including a topic sentence sequence; and determining a topic of the to-be-recognized sentence sequence according to an associated parameter, the associated parameter including the similarities between the to-be-recognized sentence sequence and the each of the topic templates in the topic template set. This embodiment reduces labor costs during a topic segmentation.