App Name Search System Using Machine Learning for Language Variations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing app name search function on smartphones is not effective, as it fails to find apps when users input partial or misspelled names, especially when dealing with loanwords and language variations, leading to inconvenient re-searching and low reliability.
Innovation Solution
A method and system that utilize a machine learning technique to process the app name and language set on a user terminal, converting and modifying the app name to match the search database, allowing for accurate app name searches across different languages and keyboard settings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the app name search function uses exact character string matching, then the search process is simple and fast, but the search accuracy deteriorates when users input partial names, misspelled names, or loanword variations
Solution Approach 1:
The system performs preliminary processing of app names by storing multiple variations (original names, loanword translations, partial names) in a database before the search operation. When an app is installed, its name is pre-processed and stored in various forms, so that during search, the system can directly compare the input against these pre-prepared variations without performing complex real-time analysis.
Solution Approach 2:
The patent introduces a database as an intermediary between the search input and the app matching process. This database stores app names in multiple formats (original Korean names, English loanword versions, partial names) and acts as a mediator that facilitates flexible matching without requiring complex real-time processing logic in the search algorithm itself.
2Adaptability or versatility
If the system performs real-time language verification and translation during search, then the adaptability to different languages improves, but the processing time and terminal load increase
Solution Approach 1:
Language translations and variations are pre-computed and stored in the database during app installation rather than being generated in real-time during search. The system prepares Korean translations, English loanword versions, and other language variations in advance, eliminating the need for real-time translation processing during user searches.
3Reliability
If the system stores multiple language variations and pronunciations of app names, then the search reliability improves, but the database size and storage requirements increase
Solution Approach 1:
The system applies local quality by storing different levels of name variations selectively based on their usefulness. Not all possible variations are stored with equal detail - the system stores essential variations (original name, key loanword translations, common partial names) with higher quality/completeness, while omitting less critical variations, thereby optimizing the balance between search reliability and database size.
Data Source
AI summary
Provided is a method of searching for an app name in a user terminal. The user terminal requests a download of an app from an app storage server. The user terminal transmits a language set in the user terminal and an app name of the app to an app name processing server. The app name processing server acquires a word in the set language corresponding to the app name through a machine learning technique, and stores the acquired word in an app search database (DB) connected in the user terminal.