Virtual Idol Conversation Processing with Vector Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing virtual idols struggle to match their responses with their attribute settings, leading to inconsistencies and a poor user experience.
Innovation Solution
A conversation information processing method that acquires attribute and conversation structures for a target object, calculates vector sets from keyword sets, generates joint semantic and structure vectors, and determines the degree of matching between the target object and conversation information to ensure consistent and authentic responses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional conversation processing methods are used, then the system is simple and easy to implement, but the matching accuracy between conversation information and virtual idol attribute settings is poor
Solution Approach 1:
The patent segments the conversation information processing into multiple independent modules: attribute structure acquisition, conversation structure acquisition, vector set calculation, joint semantic vector generation, joint structure vector generation, and matching degree determination. Each module handles a specific aspect of the processing, improving matching accuracy while keeping individual modules manageable in complexity.
Solution Approach 2:
The patent transforms conversation information and attribute settings into multi-dimensional vector representations (joint semantic vectors and joint structure vectors). This dimensional transformation enables sophisticated matching calculations by comparing vectors in a high-dimensional space, significantly improving matching accuracy beyond traditional text comparison methods.
2Measurement precision
If vector calculations and joint vector generations are performed, then the matching accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent performs preliminary actions by pre-calculating and storing attribute structures, conversation structures, and their corresponding vectors. This allows the matching process to reuse pre-computed vectors rather than recalculating them each time, reducing the computational power required during actual conversation processing while maintaining high matching accuracy.
Solution Approach 2:
The patent creates vector copies of conversation information and attribute settings, enabling efficient comparison operations. Instead of processing raw text repeatedly, the system works with copied vector representations that can be quickly compared and processed, reducing computational overhead while preserving matching accuracy.
Data Source
AI summary
A conversation information processing method, includes: acquiring an attribute structure and a conversation structure that correspond to a target object in response to detecting conversation information to be output; calculating a vector set corresponding to a keyword set; generating a joint semantic vector according to the vector set, and generating a joint structure vector according to the attribute structure and the conversation structure; and determining a degree of matching between the target object and the conversation information according to the joint semantic vector and the joint structure vector, and outputting the conversation information in response to the degree of matching meeting a preset condition.