Text Input System Using Correction Evidence for Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing text input systems, such as predictive virtual keyboards, often lead to user frustration due to incorrect autocompletion or autocorrection, requiring users to manually correct mistakes, which is burdensome and inefficient in terms of user effort and resource consumption.
Innovation Solution
A text input system that combines evidence from a user's first attempt and subsequent corrections, using an input model and text predictor to compute predicted text items, allowing for more accurate autocorrection and autocompletion, even for out-of-vocabulary words, by considering key press data and deletion events.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If predictive technology is used to autocomplete or autocorrect text items, then text input speed is improved, but accuracy deteriorates leading to incorrect text input
Solution Approach 1:
The system monitors user corrections (deletions and re-typing) and feeds this information back to the predictive model. When a user deletes predicted text and re-types, the system learns from this correction signal and adjusts future predictions, improving accuracy while maintaining speed through continued predictive functionality.
Solution Approach 2:
The predictive text system dynamically adapts its behavior based on user interactions. The model adjusts its predictions in real-time based on deletion events and re-typing patterns, making the system flexible and responsive to user intent rather than relying on static prediction rules.
2Measurement precision
If users manually correct incorrect autocompleted or autocorrected text items, then text input accuracy is improved, but user burden and time consumption increase
Solution Approach 1:
The system performs self-correction by automatically detecting when users delete predicted text and re-type it, then using this information to improve future predictions. This reduces the need for manual correction effort while maintaining high accuracy, as the system learns from user behavior patterns.
Solution Approach 2:
The patent replaces manual mechanical correction actions with an automated learning system. Instead of requiring users to manually fix each error, the system automatically processes deletion and re-typing events to improve future predictions, substituting user labor with computational learning.
3Measurement precision
If traditional predictive models are used, then common words are predicted accurately, but out-of-vocabulary words cannot be predicted
Solution Approach 1:
The system captures and stores evidence of user typing behavior (including deletions and re-typing) before final predictions are made. This preliminary evidence collection allows the system to build predictions based on actual user input patterns rather than relying solely on pre-trained vocabulary, enabling it to handle out-of-vocabulary words effectively.
Solution Approach 2:
The patent changes the parameters used for prediction from static vocabulary-based probabilities to dynamic evidence-based probabilities that incorporate user correction behavior. This allows the system to adapt to new words and phrases that users frequently type, even if they are not in the traditional vocabulary database.
Data Source
AI summary
A text input system is described for inputting text to a computing device. The text input system has a memory storing first evidence comprising text selected by a user for input to the computing device in a first attempt by a user to input intended text. The memory stores second evidence comprising either information about text deleted by the user or text selected by the user in a second attempt at inputting the intended text. The text input system has an input model configured to combine at least the first and second evidence to produce combined evidence; and a text predictor configured to take the combined evidence as input and use the combined evidence to compute a plurality of predicted text items for input to the computing device.