Entry Recommendation System Using Multi-Path Recall and Score Normalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing entry recommendation systems fail to accurately meet the input demands of users due to the limitations of system word libraries and the lack of personalization, leading to low accuracy in recommended entries.
Innovation Solution
A method and apparatus for entry recommendation that involves recalling multiple vocabularies using a multi-path recall policy, normalizing original scores to obtain target scores, determining weights for entries, and selecting target entries based on these scores and weights for push to a target client.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a system word library is used to store frequently used words, then the system can provide basic entry recommendations, but it cannot meet personalized input demands of users with emerging new entries and expressions
Solution Approach 1:
The patent segments the word library into multiple vocabularies (system word library, user word library, and hot word library) with different functions and update frequencies. This segmentation allows the system to maintain a stable core while incorporating personalized and dynamic content, resolving the contradiction between adaptability and reliability.
Solution Approach 2:
The patent implements dynamic update mechanisms where the user word library and hot word library are continuously updated based on user input behavior and trending entries, while the system word library remains relatively stable. This dynamic approach enables the system to adapt to personalized demands while maintaining overall reliability through the stable system word library.
2Productivity
If multiple vocabularies are recalled with original scores, then more entries can be selected, but the scores from different vocabularies cannot be directly compared
Solution Approach 1:
The patent applies normalization to transform original scores from different vocabularies into a unified scale, creating equipotential conditions for comparison. By mapping scores from various vocabularies to a common distribution, the system enables direct comparison and selection of entries across different sources, resolving the measurement precision issue while maintaining high productivity.
3Device complexity
If entries are directly aggregated from recall results, then the process is simple, but the recommendation accuracy is low due to lack of weighted selection
Solution Approach 1:
The patent introduces weight parameters for different vocabularies and entries within vocabularies, transforming the simple aggregation process into a weighted selection process. This parameter change allows the system to prioritize certain entries based on their importance and relevance, significantly improving recommendation accuracy while maintaining reasonable processing complexity through efficient weight calculation methods.
Data Source
AI summary
An apparatus and method for entry recommendation includes: recalling a plurality of vocabularies based on a multi-path recall policy, entries in each of the vocabularies having original scores; normalizing the original scores for the vocabulary to obtain target scores of the entries in the vocabulary; respectively determining weights of the entries in the vocabulary; selecting target entries from the plurality of vocabularies based on the target scores and the weights; and pushing the target entries to a target client.