Language Input Correction via Probabilistic Candidate Ranking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The physical limitations of keyboard interfaces on mobile devices lead to inaccuracies and inefficiencies in text input, particularly for languages that rely on phonetic-based inputs, resulting in a high number of irrelevant candidates and compromised user experience and productivity.
Innovation Solution
A system and process for language input correction that detects a sequence of contact inputs on a touch-sensitive display, determines candidate words based on the probability of intended key selections, and displays a portion of these candidates for user selection, thereby correcting unintended inputs and improving accuracy and productivity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a phonetic-based input system is used on a mobile device keyboard interface, then text input capability is enabled, but input accuracy deteriorates due to physical limitations of the keyboard size
Solution Approach 1:
The system detects contact inputs on the keyboard interface, determines candidate words based on phonetic conversion, ranks candidates by probability, and displays them for user selection. This feedback loop allows the system to learn from user selections and improve accuracy over time while maintaining ease of operation through phonetic input
Solution Approach 2:
The patent introduces an intermediary processing layer that converts phonetic inputs into predicted candidate text before final selection. This intermediary step acts as a buffer between the user's finger input and the final character selection, allowing for error correction and improving overall input accuracy without requiring larger keyboard keys
2Volume of moving object
If the keyboard interface size is reduced to fit mobile devices, then device portability is improved, but touch input precision deteriorates
Solution Approach 1:
The patent replaces the mechanical precision requirement of physically larger keys with a computational system that uses probability ranking and candidate display. Instead of relying on mechanical key size for precision, the system uses software-based phonetic conversion and probabilistic candidate ranking to achieve accurate text input on small mobile device keyboards
3Productivity
If phonetic-based input conversion is performed, then text generation capability is improved, but candidate relevance deteriorates due to error propagation
Solution Approach 1:
The system performs preliminary phonetic conversion of contact inputs into predicted candidate text before final selection. By converting phonetic inputs to candidate words in advance and ranking them by probability, the system prepares multiple likely options for the user, improving text generation capability while maintaining candidate relevance through probabilistic filtering
Solution Approach 2:
The candidate selection process provides feedback to the phonetic conversion system. User selections from the displayed candidates inform the system about actual intent versus predicted intent, allowing the system to refine its phonetic conversion and candidate ranking algorithms to improve both productivity and candidate relevance over time
Data Source
AI summary
The present disclosure generally relates to language input correction. In one example process, a sequence of contact inputs can be detected via a keyboard interface on a touch-sensitive display. A contact input of the sequence of contact inputs can include a contact motion from a first position to a second position of the keyboard interface. A plurality of candidate words corresponding to the sequence of contact inputs can be determined. The plurality of candidate words can be ranked based on a probability that the contact input is an intended input to select a first key of the keyboard interface, and a probability that the contact input is an intended input to select a second key of the keyboard interface. A portion of the plurality of candidate words can be displayed for user selection.