Query Prefix Perplexity Filtering for Data Prefetching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing prefetching request mechanism based on query prefix completion and matching has a low success rate, leading to system instability and increased machine costs due to excessive requests, which negatively impacts user experience.
Innovation Solution
A data prefetching method that analyzes the query prefix using a pre-trained linguistic model to determine its perplexity, filtering out high-perplexity requests by transmitting prefetching requests only when the perplexity is below a predetermined threshold, and optionally applying a correction factor for special characters to optimize perplexity calculation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If prefetching requests are issued based on query prefix completion and matching, then the searching speed is improved, but the prefetching success rate is low and system stability is adversely affected
Solution Approach 1:
The patent changes the parameter used for filtering prefetching requests from simple prefix matching to perplexity score thresholding. By calculating the perplexity of each query prefix and comparing it against a predetermined threshold, the system selectively issues prefetching requests only for high-quality prefixes, thereby improving the prefetching success rate while maintaining search speed.
2Speed
If prefetching requests are issued based on query prefix completion and matching, then the searching speed is improved, but machine cost increases due to excessive requests
Solution Approach 1:
The patent introduces a perplexity threshold parameter to control the issuance of prefetching requests. By filtering out low-perplexity requests, the system reduces the total number of prefetching requests sent to the backend server, thereby reducing machine cost and energy consumption while preserving the speed benefit of prefetching for high-quality queries.
3Speed
If prefetching requests are issued based on query prefix completion and matching, then the searching speed is improved, but system stability is adversely affected
Solution Approach 1:
The patent applies perplexity thresholding to filter prefetching requests, which stabilizes the system by preventing the processing of low-quality or malformed queries. This parameter-based filtering ensures that only meaningful query prefixes trigger prefetching operations, reducing system instability caused by excessive or inappropriate requests.
4Speed
If prefetching requests are issued based on query prefix completion and matching, then the searching speed is improved, but user visual experience is interfered due to excessive prefetching results
Solution Approach 1:
The patent uses perplexity thresholding to control the generation of prefetching results. By filtering out low-perplexity query prefixes, the system reduces the number of prefetching results displayed to users, eliminating visual clutter and improving user experience while maintaining the speed advantage of prefetching for high-quality queries.
Data Source
AI summary
The present disclosure provides a data prefetching method, a data prefetching apparatus, an electronic device and a computer-readable storage medium. The data prefetching method includes: acquiring a query prefix inputted by a user; analyzing the query prefix in accordance with a pre-trained linguistic model to acquire a perplexity of the query prefix; judging whether the perplexity is smaller than a predetermined threshold; and transmitting a prefetching request in accordance with the query prefix when the perplexity is smaller than the predetermined threshold.